Assistive Robotics

The following sections are included:

• PREFACE
• CONFERENCE ORGANISERS
• CONFERENCE COMMITTEES AND CHAIRS
• CONFERENCE SPONSORS AND CO-SPONSORS
• TABLE OF CONTENTS

Current applications of robotics is distinguished from more traditional automation by the focus on machines that operate in relatively unstructured, difficult and often hazardous environments. The past decade has seen the deployment of a number of robotic systems in highly challenging application domains such as mining and agriculture. In particular, the potential safety, cost and health impacts from the use of robotics aids for periodic inspection and maintenance of civil infrastructure has led to a significant expansion of research activity in “infrastructure robotics”. Two of the prerequisites for deploying a robot in the field are the ability to acquire and maintain a representation of the environment, and efficient motion planning. In scenarios where a machine and a human must collaborate to perform a task, an intuitive user interface, a joint understanding of abilities and joint management of task execution are also essential. This talk will summarise current research on these key competencies that underpin robot deployments in unstructured environments, with the focus on simultaneous localisation and mapping, and human robot collaboration. Future potential and key challenges in infrastructure robotics will also be presented together with a review of the research activities in this field at the Centre for Autonomous Systems, University of Technology Sydney.

In a dynamic and changing world, a robust and effective robot system must have adaptive behaviours, incrementally learnable skills and a high-level conceptual understanding of the world it inhabits, as well as planning capabilities for autonomous operations. Future robot systems will benefit from the recent research on neurocognitive models in processing multisensory data, exploiting synergy, integrating high-level knowledge and learning, etc. I will first introduce multisensory integration methods for intelligent control of robots. Then I will present our investigation and experiments on synergy technique which uses fewer parameters to govern the high DOF of multifinger robot movement. The third part of my talk will demonstrate how an intelligent system like a robot can evolve its model as a result of learning from experiences; and how such a model allows a robot to better understand new situations by integration of knowledge, planning and learning. I will show some integrated results of operational mobile robot platforms with grasping facilities in a restaurant service scenario.

Robots will transform our daily lives in the near future by moving from controlled industrial lines to unstructured environments such as home, offices, or outdoors with various applications from healthcare, service, to defence. Nevertheless, two fundamental problems remain unsolved for robots to work in such environments. On one hand, how to equip robots, especially meso-scale ones with sizes of a few centimetres, with multiple locomotion abilities to deal with the unstructured environment is still a daunting task. On the other hand, how to control such robots to dynamically interact with the uncertain environment for agile and robust locomotion also requires tremendous efforts. In this talk, I will present my research efforts to tackle these two problems in the framework of biologically inspired robotics. First, I will show how to use biologically principles found in nature to build efficient meso-scale robots with various locomotion abilities such as jumping, rolling, and aerial manoeuvring. Second, I will present a novel non-vector space control method for meso-scale robots which have limited computation power. The research in these two thrusts will pave the way for next generation bio-inspired, low cost, and agile robots.

The current studies of fish-like aquatic propulsion are quite limited in their ability to control for physical determinates such as body shape, kinetic motions, flexural stiffness, skin surface scale and Mucus properties etc. Keeping one of these determinates constant while altering others in a controllable manner is impossible for the swimming live fishes. Bio-inspired robotic models offered the ability to manipulate and control individual physical determinate that affect the aquatic propulsive performance…

This paper proposes a behavior adaptation method for companion robot based on the hierarchical partially observable Markov decision processes (H-POMDPs). The HPOMDPs have recently emerged as a powerful method for optimizing human-robot interaction (HRI) decision making in partially observable environments. The model is used for representing and learning HRI models for indoor elderly companion robot. This approach proposes an approximate solution by employing the H-POMDPs in the uncertain environments. The HRI states such as daily dialogues, news and whether broadcasting, motion speed and navigation-assist are used through the H-POMDP. The user's speech command, touch-screen input, head position and body posture are detected as subconscious signals that indicate a user's interaction preference. The experimental results show the effectiveness and feasibility of designing the H-POMDP behavior adaptation method. The H-POMDPs can train significantly faster than the original MDP. The H-POMDPs requires much less data, and can easily be extended to variables reducing the time and sample complexity.

Robotic transtibial prostheses which can intelligently adjust ankle angle are gaining popularity. In this paper, based on the light-weight robotic transtibial prosthesis with damping control, we further study the improvement on amputee's walking performance at different speeds on level-ground. Experiments on one transtibial amputee subject indicate that, compared with the passive prosthesis, the proposed robotic transtibial prosthesis has a larger range of ankle movement, which is closer to that of the able-bodied, and can improve amputee's gait symmetry and walking stability at different speeds.

Following common mechatronic design methodology VDI 2206, this paper introduces three different variants of wheg design. During construction focus was placed on a smooth locomotion of all wheg-mechanisms over flat terrain, which means low alternation of CoM in vertical direction y (in the following addressed as ‘quietness’ of a wheg) as well as the ability to overcome obstacles. Subject to the number of spokes, quietness and theoretically manageable obstacle height were calculated. The results were taken as reference for experimental evaluation of each variant and a successive comparison. The phase shifting double wheg, is described in further detail.

This paper proposes a design of above knee prosthesis that allows above knee amputees to walk on ground level and ascend/descend stairs by using only hydraulic circulation system (HCS) as self-energized actuator. The mechanism utilizes amputee thigh motion to drive oil flow in the HCS to enable the prosthesis to make extension, flexion, dorsiflexion and plantar flexion. Oil flow is closed for ascending stairs and adjusted for level walking and stairs descending by a flow control valve (FCL).

The design of a controller is one of the key tasks and major difficulties in the development of rehabilitation exoskeletons. An algorithm about an adaptive gait trajectory based on the iterative learning control for lower extremity rehabilitation exoskeleton is proposed in this paper. First of all, dynamic model is built up based on the Lagrange equations for the lower extremity rehabilitation exoskeleton. Secondly, an adaptive gait trajectory based on the iterative learning controller is put forward to achieve the active mode of patients. Finally, a simulation experiment is conducted in MATLAB based on the standard gait data which were collected by an optical motion capture system. The simulation results show that the control algorithm can achieve the desired adaptive tracking for joint trajectory and enable patients' active participation in rehabilitation.

In recent years, due to the development of computers and the Internet, the robotics industry has steadily grown. Besides academic applications, entertainment robots can also be programed to bring us unprecedented performances that many people can enjoy. Robotic puppet shows are an important part of the entertainment robot field. The robot can integrate the basic ability to speak, make poses and human like expressions. Based on the considerations above, this paper describes an Editorial Platform for Screenplay of Interactive Robotic Puppet Shows (EPFS). EPFS lets users in different age groups control robots and allows them to write the screenplays for drama performances in a simple and fast way. In the platform users have the ability to use game design concepts that allow them to make diverse screenplays and a multitude of different storylines. Finally, the platform controls and records the story-making process by constructing a timeline. In doing so, problems caused by robotic expression and interaction can be solved. In addition, the arrangement of screenplays will be more flexible and accurate. According to the results of the experiment the platform proves to be flexible and easy to use. A video presentation of the paper has been posted on YouTube; please search the title below “An Editorial Platform for Screenplay of Interactive Robotic”.

The paper describes the design details for realising the EXO-LEGS assistive exoskeletons for Ambient Assisted Living (AAL) applications based on modelling and simulation studies performed for key mobility functionalities in activities for daily living such as stable standing in open space and straight walking. The results provide the basis for selecting sensors and actuators to develop the needed assistive exoskeletons to help the elderly to stay active and independent for as long as possible.

This paper presents investigations for development of an assistive exoskeleton device for elderly mobility. This exoskeleton is designed to enhance the lower limb and provide support torque in order to augment the torque of knee and hip during the walking cycle. PID Control is designed and implemented in this work. Due to the complexity in identifying the lower limb musculoskeletal system with traditional mathematical approaches, the visual Nastran 4D software is used for development of simulation model of the exoskeleton and a humanoid. Simulation results demonstrating the performance of the adopted approach are presented and discussed.

Crutch and walking frames are widely used together with exoskeleton assistive devices to provide body weight support and balance to the user, However, it is unclear what is an optimal posture for the user to choose when using these devices, In this paper, it is found out that human posture, the force from the upper body and the support points are the three key effectors that affect the stability during assisted walking. And it is proposed to evaluate the effect of human posture through a global index which is widely used in robotics field: the manipulability. A three degree of freedom arm model is used and an optimization framework is established to find out the optimal posture during assistive walking.

This work presented in this paper focuses on development of assistive robotic control approach for the upper extremities. A set-point tracking position control structure with proportional, integral and derivative (PID) controller is considered. A spiral dynamic optimization algorithm is utilized for tuning of the PID gains. The control strategy is tested and evaluated within simulation model of upper extremities. The results show that good position tracking performance is achieved with the developed control approach.

This paper presents research on the modelling and simulation of assistive exoskeletons for elderly mobility. The exoskeleton is designed to augment physical capabilities of the elderly, so that the combined natural joint torques, and torques provided by the exoskeleton, result in overall torque equivalent to that of a physically able person. The main mobility functions considered are standing-up, sitting-down, upright balance and level walking. The current paper focuses on standing-up and sitting-down scenarios. Torque profiles required for achieving desired joint displacement are investigated to establish upper limits for control actuation applied at the exoskeleton joints to ensure smooth manoeuvres.

The present paper aims to contribute to the definition of the structure and substance of long-run experiences indispensable to assess the nature and quality of Human-Robot Interaction (HRI) by describing the procedures involved in the definition of those experiments in the context of the MOnarCH project.

This paper describes an artificial potential field algorithm for the path-finding problem solving in dynamically changing environments for a wall-climbing robot. Suitability of solving path-finding problem with artificial potential field approach for the wall climbing robot with sliding seal has been analyzed. Simulation demonstrated that this approach provides real-time creation of accurate and smooth trajectories.

One of the central problems of motion control of walking robots is the distribution of force between legs and the organization of robot motion within margins of static stability. Support reactions need to be controlled during movements on undeveloped terrain. The force/torque sensor is an important component of the measuring system and force control and sliding contact with the ground uneven leg of the modular walking robot MERO. Determination of the real forces distribution in the shifting mechanisms of a walking locomotion system which moves in undeveloped terrain at low speed is necessary for the analysis of stability. The paper will analyze an intelligent sensor developed by the authors with an important role in the complex system control slipping when moving walking robot MERO on undeveloped terrain.

This paper presents a general framework for an autonomous climbing robot. The objective of the work is to design climbing robot by using robot arm which has the capabilities to grasp the surface, i.e. pole. The robot consists of two arms with wheels attached to each arm enabling the robot to climb or descend a steep surface such as poles or gap between walls. The arms are used for grasping to the surface while the wheels are used for movement upward or downward. The wheels offer friction contacts which allow to apply force on the surface for stable climbing and descending. Fundamental challenges to the development of real robotic systems include the hardware design, control system, grasping and manipulation technique and analysis of the scansorial robot.

Simultaneous Localization and Mapping (SLAM) algorithms with autonomous robots have received considerable attention in recent years. In general, those algorithms use odometry information and measurements from exteroceptive sensors of robots. The accuracy of the map building and the robot localization algorithms directly affect the overall success of the system. This paper proposed a novel method in map building and robot localization. Unlike traditional approach in building maps, which uses the recursive algorithm and is of low efficiency and occupies much memory, a gradient based algorithm is proposed which has the advantages of concise form, high efficiency and accuracy. Several experiments are taken to demonstrate the comparisons between the ordinary method and the novel one. In localization, Kalman Filter (KF) can evaluate the state of the system from the dynamic circumstances, but it must obey the rule that the noise is of Gaussian distribution, so the modified form (EKF) is adopted on many occasions. Compared to other algorithms such as Markov localization, Mont Carlo localization, EKF is more accurate. By taking the experiments, the applicability of the algorithm is verified. All these techniques have been implemented on mobile robot Pioneer P3-AT equipped with a 2D laser range scanner SICK LMS 200.

Though over decades' development of bipedal robots, the terrains that bipedal robots can walk remains limited compared to what human can accomplish. To increase bipedal transversality on upslope terrain, this paper studies the biomechanical and biological aspects of human walking on inclined slopes with underlying the motor skills and reflexive systems. The control strategy can be divided into low and high gradient upslope walking. The strategy for the low gradient uphill walking is generated on the basis of an existing B4LC system. Furthermore, investigating the human walking on high gradient upslope terrain thoroughly unveils a new control strategy for bipedal walking on high gradient slope. Through validating the suggested method on a simulated biped upon different upslope terrains, the bipedal robot shows a naturally looking walking gait, achieving uphill walking up to 15° inclination which can compete with most advanced bipedal robots in the world.

Walking on uphill sandy surface is a challenging task and is considered one of the most difficult terrains that a walking robot can face. This paper will shed light on our applied decentralized approach to enable a hexapod robot to walk on uphill sandy surface effectively. In this paper we will show that the combination between our previously introduced approaches and our new strategy enables the hexapod robot to cope with this type of terrain. The suggested strategy is based on the synchronization between the moved legs that are touching the ground. This synchronization provides the robot's legs with the sufficient driving force during uphill walking. The novelty of our approach is the only evaluation of the local current consumption and angular position of each leg's joint as somatosensory feedback. It is based on an organic computing architecture and was tested on a low-cost version of the OSCAR walking robot.

Nature has adapted to allow the survival of its species through the development of several forms, materials, techniques and processes. These features, such as flying, defensive and offensive mechanisms and many others have inspired researchers, creative and enthusiasts into developing artificial solutions that mimic their natural counterparts. Early developments of biomimetic involved large structures, however, with current advances in mechanics, electronics, and computer science the development of micro and nano structures is allowing the creation of more complex, flexible and portable structures resembling biological features found in nature. These advances in robotics may help solve or improve tasks otherwise difficult with traditional robotics. This paper focuses on the design and development of soft-robotic manta ray prototype whose motion is controlled using magnetic fields over a body composed of a ferro-fluid. The goal is to explore magnetics actuation as an alternate mean to control and reduce soft-robotics size as an alternative to current traditional mechanical or electromechanical biomimetic developments and even shape memory alloys.

The article presents the mathematical model of the anthropomorphic robot, which rises from an immovable surface, the stage-by-stage algorithm of the mechanism operation, the results of the numerical modeling, the dependence of the change torques generated by the actuators from the laws of the rotation angles of the mechanism links.

A baby-mimic quadruped robot with global flexible spine of variable cross-section stiffness mechanism scheme is proposed aiming at solving the problem of unstable body posture when the quadruped robot is walking. The trot gait of ‘BabyBot’ is realized by adopting center pattern generator method based on biological neural control mechanism. The dynamic simulation analysis is carried out and the experimental platform is set up. In addition, the walking experiments and performance evaluations of the baby-mimic quadruped robot are conducted. Experiments verify that adding the baby-mimic quadruped robot with a global flexible spine of variable cross-section stiffness has a positive effect on reducing the swing of the body posture (head and hind torso). Thereby, it is effective for improving walking coordination and posture stability for legged robots.

In this study, we analyzed the bionic structure of cockroach legs and observed the morphological characteristics of the hooks on cockroach legs in reference to structural bionic principles. We investigated the interaction mechanism of the bionic hook with the bulges on rough surfaces and deduced the mechanical conditions of the hook stably grabbing the wall bulges. Then, we presented an initial design of a wall-climbing robot based on a grabbing claw and proposed a climbing model. Furthermore, we performed several experiments to demonstrate the grasping stability. The grasping claw of the climbing robot can offer a more reliable way of adsorption for complicated rough surfaces.

Smart intelligent robotic fish has shown promising advantage in underwater searching. This paper addresses the smart robotic shark design and control issues with multi-sensors. In particular, we propose a new design of a two-link mechanism robotic shark equipped with gyroscope, pressure sensor, infrared sensor, and light sensor. Then three-dimensional motion control, depth control, autonomous obstacle avoidance, and light navigation are developed. In particular, a bio-inspired Central Pattern Generator (CPG) based control method is adopted to smoothly control the robotic shark's locomotion in all the above realization. All motion control methods are implemented in real time with a hybrid control system based on embedded microprocessor (STMicroelectronics STM32F407). Latest aquatic experiments demonstrate a fairly good result in improving the robotic shark's intelligence. The developed scheme affords an alternative to smart robotic fish design in complex underwater environments.

Dynamic movement primitives provides an approach to generate and modulate periodic trajectories real time. Based on DMP, We proposed an open loop position control quadruped controller which is able to make the robot trot on a flat plane. In this modular controller, each Single leg module is responsible for one leg motion generation, and this motion generation is interpreted in Cartesian space. Such design makes it easy to modulate the four legs phase relation and independent of mechanical structure. Speed control is achieved through frequency control which even allows moving backwards without a stop smoothly. Turning control is achieved through amplitude control. Our controller is easy to apply and expand but quite effective of phase coordination and online trajectory modulation. Results of simulated experiments on moving, turning and speed control are presented.

This paper presented a tortoise-like robot with various gaits in different terrestrial environment. Firstly, the locomotion mode of tortoise was particularly investigated, including triangular gait, diagonal gait and synchronous gait. The trials for acquiring the trajectory of tortoise's limb prepared principal information for the fabrication of robot's limb, structural design and corresponding control. Secondly, the exploration of soft limbs presented an efficient method to develop new bionic robot. The optimization of structure and the driving system help to reduce the number of motors for the robotic system. Finally, the robot prototype is able to realize multi-gaits driven by crank shaft-rope actuation system via sequence control method. The results of terrestrial locomotion experiments verify the adaptability of the robot in different environmental terrains.

Minimalistic gait models assuming massless legs (e.g. models based on spring loaded inverted pendulum (SLIP) or inverted pendulum (IP) template) have been widely used to interpret human gaits. Although these models can describe basic gait features like center of mass (CoM) trajectories or ground reaction force (GRF) patterns, it is rather challenging to investigate swing leg control strategies with these massless leg models. In this paper, based on experiment data, we analyse how much swing leg, stance leg and upper body movement contribute to total GRF during single-support phase of walking. The results show that, in vertical direction, swing leg and upper body create in-phase M-shape force patterns, but stance leg does not contribute to the M-shape force pattern. In walking direction, the inertia forces created by swing and stance leg cancel each other out, while stance leg and upper body create similar inertia forces in both shape and magnitude. The results suggest there is a phase locking mechanism for swing leg, stance leg and upper body movement. It can help to refine current conceptual models which better describe human walking.

Bio-robotic models become increasingly important for understanding biological system in field such as biomechanics. Fish caudal fin is a prominent example of biological propulsion, in which the caudal peduncle, fin ray and fin membrane together form a dynamic locomotory system. In this paper, we developed a bio-robotic model to mimic the fin ray structure and kinematics of Bluegill Sunfish (Lepomis macrochirus). We coupled controlled oscillations in both heave and pitch directions to the robot to model the caudal peduncle motion of swimming fishes. Synchronized multi-axis force transducer and particle image velocimetry were then used to quantify the hydrodynamic forces and wake flow. We found that the addition of three-dimensional fin kinematics significantly enhanced the lift force without deceasing thrust force compared with the no fin motion. The vortex wake directs water both axially and vertically and forms jet like structure with notable wake velocity. According to the bio-robotic model experimental data, we hypothesized that fish may actively control the caudal fin rays to achieve considerable lift force when swimming at low speed, however, negative at high speed.

This paper concentrates on hysteresis nonlinearity that comprehensively exists in bio-inspired locomotion of clawing, walking, running, flying and/or swimming robots. A memory clearing iterative learning control scheme is employed into hysteresis-compensating. A second-order state-space model with Preisach-type hysteresis is addressed to describe the dynamic features of biomimetic joints and components. The memory clearing operator enables feasibility of iterative learning control (ILC) methods, whether the state initial errors are accurately set or not. The simulation examples validate that combining a pre-operator with an iterative controller can effectively eliminate effects caused by the Preisach-type hysteresis in bio-inspired robotic systems.

This work explores the use of active tails for steady-state legged-locomotion. Simple models are proposed which capture the dynamics of an idealized running system with an active tail. Analysis suggests that the control objectives of injecting energy into the system and stabilizing body-pitch can be effectively decoupled via proper tail design: a long, light tail. Thus the overall control problem can be simplified, using the tail exclusively to stabilize body-pitch. We show in simulation that models with long-light tails are better able to reject perturbations to body-pitching than short-heavy tails with the same moment of inertia. We also present the results of an active tail mounted on the quadruped robot Cheetah-Cub. The results show greatly improved forward velocity and reduced body-pitching and validate the long-light tail design: shorter, heavier tails are much more sensitive to control parameter changes.

This article is devoted to the kinematics of the multilink body of a multilegged walking robot. The number of degrees of freedom for multilink body is obtained and the reverse kinematic problem is solved. The control algorithm for mosaic robot body is provided. All algorithms have been verified via multibody computer simulation.

Regarding the working task requirement, a novel robot mechanism for inspecting power transmission lines with multi-unit mechanisms in series is designed. Several key mechanical design features are briefly described. The dynamics and stiffness characteristics of the parallelogram structure (PS) driven by two flexible cables are researched. The analysis results show that the stiffness of the robot in vertical plane can be controlled by actively changing elevation angle and cable tensile force of PSs.

A common disadvantage of quadruped robots is that they are often limited to load carrying or observation tasks, due to their lack of manipulation capability. To remove this limitation, arms can be added to the body of the robot, enabling manipulation and providing assistance to the robot during body stabilization. However, a suitable arm for a quadruped platform requires specific features which might not all be available in off-the-shelf manipulators (e.g. speed, torque-controlled, light-weight, compact, without external control unit). In this paper, we present a systematic approach to design a robotic arm tailored for an 80kg quadruped robot. A full robot with arms and legs (aiming for a centaur-style robot) was simulated performing a range of “representative” tasks to estimate joint torques and velocities. This data was then extensively used to select the design parameters, such as the joint actuators to develop a novel, compact (0.743m fully extended), light-weight (12.5kg), and fast (maximum 4m/s no-load speed at end-effector) hydraulically actuated robotic arm with 6 torque-controlled degrees of freedom. The enclosed video presents preliminary experimental results.

Many robotic systems use similar sensors and actuators and navigation and localization algorithms. The adoption of a modular approach can greatly improve reusability of sub-systems and helps team work and system maintenance. Graphical programming environments can furthermore support this process, avoiding writing textual code, by using simple graphical objects. In this paper we show three different experiences we had by using a modular approach in the software programming of three different robotic systems. LabVIEW software environment was used and its main advantages will be briefly described.

This paper presents Rise-Rover, a new generation wall-climbing robot with high reliability and load-carrying capacity on vertical surfaces for Non-Destructive Testing (NDT) of concrete and steel infrastructure. One Rise-Rover drivetrain module can operate on both smooth and rough vertical/inclined surfaces independently. The on-board electronics and PID controller monitor the pressure reading and adjust impeller speed to provide stable suction force for the wall-climbing robot. The use of duct fan and tether further increase the operation reliability. Rise-Rover is remotely controlled by Android smartphone via Wifi, and the User Interface (UI) provides good usability and convenience. The experimental test verified the good performance of the Rise-Rover prototype.

High-performance legged robots that are required to navigate on unstructured and challenging terrain benefit from torque-controlled joints. High-fidelity torque measurements are crucial for proper joint torque control. Commercially available torque sensors are expensive and often hard to integrate into compact and light-weight robot leg designs. Custom-made sensors on the other hand often suffer from asymmetric behaviour with respect to direction of rotation or poor linearity, especially for small and compact applications. This work is motivated by the need to achieve reliable torque measurements for the newly developed, small-size hydraulically actuated quadruped robot MiniHyQ. The main contribution of this work is the development of a new innovative design of a strain gauge based torque sensor with a high degree of linearity, symmetry, and scalability (both in dimension and measuring range). Furthermore, the glueing and wiring of the strain gauges are easy thanks to the geometry of the sensor that allows direct access to the mounting surfaces, even in compact dimensions. We show the design's symmetric (clockwise and counterclockwise rotation) and linear behaviour through virtual prototyping and experimental tests. Furthermore, we show how a small-scale instance of the sensor design is successfully installed on the MiniHyQ robot.

Landmines, cluster munitions and explosive remnants of war still pose a threat to the physical, psychological, economic and social welfare of many peoples. With the aim of contributing to alleviate this terrible problem, this paper introduces a robotic manipulator capable of handling a metal detector while scans the surface of interest searching for mines during humanitarian demining tasks. A set of mini-cameras installed on the robotic arm acquires the data required for adapting the scanning to the irregularities of the terrain in such a way that the metal detector remains at a constant height from the ground during all the operating procedures. Results of preliminary tests show the potential of the proposed approach.

This paper presents an approach to using noisy and incomplete depth-camera datasets to detect reliable surface features for use in map construction for a caterpillar-inspired climbing robot. The approach uses a combination of plane extraction, clustering and template matching techniques to infer from the restricted dataset a usable map. This approach has been tested in both laboratory and real-world steel bridge tunnel datasets generated by a climbing robot, with the results showing that the generated maps are accurate enough for use in localisation and step trajectory planning.

Metastable walking is a relatively new concept in the field of walking robots. According to the theory, a walking machine is inherently subjected to stochastic disturbances due to the uncertainties about the terrain. As a result, it cannot demonstrate a limit cycle walking; but, it is still able to exhibit remarkably long-living periods of continuous walking. In this paper, metastable walking over irregular terrains has been revisited. By introducing the concept of ‘local slopes’ and highlighting their role in limit cycle walking stability, we have shown how this concept contributes to the study of metastable walking over an uneven terrain. As an instance, a simulation study is performed to demonstrate that metastable walking of a simple passive dynamic walker is achievable on irregular surfaces with the average slopes beyond the maximum slope of a simple ramp which provides the biped with a stable periodic walking. This would help to reduce the energy inputs applied to control bipedal robots on uneven terrains.

As humanoids are expected acting in the real world to humanly and intelligently complete some high-level tasks, precisely perceiving the changes of environments is thus an essential premise. Due to the complexity of real-world environments, how to exactly perceive the environmental changes with limited sensors equipped on the robot becomes a challenging problem. Though this problem can be solved by establishing direct sensory mappings or employing probabilistic filtering methods, the nonlinearity and uncertainty caused by the high degree-of-freedom of the robots and the complexity of the environment result in tough modeling difficulties. In this paper, an alternative learning approach for addressing such modeling problems under the Gaussian process regression framework is proposed. The evaluations under perceptual tasks for two representative environmental changes, i.e. unknown pushing and sloped terrains, reveal the potential of the proposed approach in coping with the nonlinearity and uncertainty of environmental perception for humanoids.

This paper presents an optimization mechanism for feed-forward control units termed as motor patterns in the Bio-inspired Behavior-Based Bipedal Locomotion Control (B4LC) system. In specific locomotion phases, motor patterns ActiveHipSwing and LegPropel produce torques at the hip and ankle joints respectively. Biped is activated to swing leg ahead and push body forward by corresponding motor patterns, in which the parameters determine the profile of generated torques according to the sigmoid function. To obtain efficient and stable locomotion control, we employ Particle Swarm Optimization (PSO) method to tune motor patterns' parameters by formulating locomotion stability, energy consumption and walking speed as fitness functions. The optimization procedure takes place on a 3-dimensional simulated bipedal robot. Simulation results prove that suggested approach reinforces the walking behavior of biped with respect to stability, velocity control as well as enhances the energy efficiency significantly.

In order to study the key factors that influence the balance of a hopping one-legged robot, we decompose the hopping balance model into two parts: leg swing control in flight phase and stability control in stance. During flight phase, since the horizontal velocity is not easy to measure by sensors and might cause cumulative error and time delay, we use body attitude and angular velocity as approximate input parameters to replace horizontal velocity in leg swing algorithm. In the experiments of stance phase, we discover the close relation between duty factor and spring rate as well as state transition threshold of pneumatic cylinder on the leg. We combine the two experiments and preliminary implemented the balance of one-legged robot hopping in place. The experiments demonstrated that body attitude and angle velocity in flight phase and duty factor in stance phase are the key factors for balance. Comparing to works made by Raibert[2] in balance controlling under high translation speed, our work provides references to study balance of robot while hopping under low horizontal velocity.

Coupled elastic actuation has been proven to be a simple and effective method to realize natural biped walking. However, former researches found it hard to achieve high speed merely by adjusting control parameters. For the purpose of realizing fast walking in this method, we put forward a simple approach: moving CoM(center of mass) forward. Through numerical simulation, we found that a positive lateral offset of CoM could effectively increase walking speed by enlarging step length and shortening step period, both of which result from the reduction of swing leg retraction. To testify the simulation results, experiments based on a prototype were conducted. The dimensionless speed of the robot changes from 0.264 to 0.424 as CoM moves forward, which confirms our speed-increasing strategy.

The present paper discusses a control method for the location (x,y) and yaw angle of the body of a six-legged robot using a laser range finder. During walking, the location and yaw angle of the body are calculated by referring to these two fixed points obtained by the laser range finder. In the experiments, the validity of the proposed walking directional control method was examined for straight walking.

In recent years, several studies have suggested that improved performance of modern robots can arise from encoding commands in terms of motor primitives. In this context, dynamic movement primitives (DMP) is a powerful tool for motion planning based on demonstrations, well-suited for robot learning. In this work, we study on-line adaptation of biped locomotion patterns when employing DMP. The goal is to demonstrate and evaluate how new movements can be generated by simply modifying the parameters of rhythmic DMP learned in task space. This formulation allows recreating new movements such that the DMP's parameters directly relate to task variables, such as step length, hip height, foot clearance and forward velocity. Several experiments are conducted using the V-REP simulator, including the adaptation of the robot's gait pattern to irregularities on the ground surface and stepping over obstacles.

This paper deals with locomotion cost (energy consumption per unit distance) in the trotting gait of our four-legged robot compared to that of horses. First, we designed a new toe with a spring to absorb the impacts at touchdown, which we expect to increase the stability of the robot. We attached the toe to the robot leg and achieved its trotting gait without any support. We measured the locomotion cost in the trotting gait of the robot at several speeds. Based on the results, we found that the locomotion cost decreases when the locomotion speed increases. We concluded that this phenomena is caused by a decrease in the number of steps.

In the applications of simulator, accurate machining and soldering, a parallel hexapod can usually be found. However, it is still an innovative research object to implement this structure in a personal biped walking application. This article represents a realization of a walking mechanism implying parallel Hexapods as moving parts. Working principle and control structure are described in this work. To realize a stable and riding comfortable walking mechanism, the planning and control of stability is fatal. This aspect is inevitable tightly related to the pose recognition algorithm. To solve a forward kinematic problem (FKP) is always a difficulty in the recognition of pose. In this article, a method integrating MEMS inertial measurement units and lengths measurements is designed. With the force sensors, the center of pressure can also be confirmed. This fusion of sensory system is further to realize a steady and agile control of this biped walking robot.

In this paper, we aim at realizing stilt-like biped walking. With the assumption that the Jacobian of task space is full row rank during the single support phase, the dynamics in task space is presented with the Moore-Penrose pseudo-inverse. Considering the inevitable parameter estimation errors, we design a task space controller without calculating the inversion of inertial matrix. We demonstrate the stability of the controller by Lyapunov method. With the designed the controller, we realized stilt-like biped walking on a virtual planar biped walker. The contact consistency is preserved during the biped walking.

This paper presents the dynamic walking of the bipedal robot, in which the long double support phase is used for compensation of the external disturbances (including contact with support surface) for stabilization of the step cycle. Motions of the robot by the single support phase are based on free (ballistic) trajectories of the mass center (ZMP is located under the foot). During the double support phase robot first slows down due the damping features of the control system, then it stops and then accelerates again to the ballistic trajectories. Experiments were performed using a robot “ROTTO” [3].

Robots are faced with increasingly complex, hard to simulate environments. Multilegged robots exhibit dynamics that are often difficult to model well. Their development thus depends on experimental verification. Here we propose “focused modularity” – an approach combining benefits of both modular robotics and more traditional rapid fabrication.We made robot mechanisms by laser cutting low-cost foam board, while the complexity of motor control and electronics is focused in a small number of servomotor modules. This dramatically reduced the cost and time of design iteration, and allowed incremental improvements to make up for deficits in material strength and manufacturing tolerance. We present meter-scale hexapods whose chassis and drive train were manufactured for under 20 USD and under four hours of skilled labor. To date we have produced over 40 revisions of the design, including 25 iterations of drive mechanism design and 7 full robot prototypes built. This represents a substantially more thorough exploration of the design space than could be possible using conventional development approaches. We suggest a library of suitable robot designs to be developed, permitting a robot to be built on the fly with functionality for a particular task as the situation demands.

In this paper, we focus on the problems of gait planning and stabilization for a three-link planar biped robot with periodic gaits in level ground. The key in designing this under actuated bipedal robot is to construct a feedback controller to realize stable locomotion. To achieve the control objectives, a time independent periodic trajectory based on the dynamic equations is designed for gait planning, and a nonlinear control including adaptive control, force control and feedback sliding mode control is designed to realize gait stabilization. In this paper, the simulation results on gait generation and stabilization are presented.

The inspection platform that is described in this manuscript consists of a hexapod walking robot designed by the Centre for Automation and Robotics CSIC-UPM, Spain. This inspection platform will load a scanning manipulator arm which, in turn, carries a metal detector on its tool centre point. With the integration of both, hexapod robot and scanning manipulator, several test tasks about the search and localisation of antipersonnel mines will be carried out, within a controlled environment. The SCARA configuration of the hexapod robot legs will allow low energy consumption when the robot executes gaits on flat terrain or with reduced slope, due the decoupling of gravitational effects. This legged robot has a mass about 250 kg, and it can bear a high payload up to about 300 kg. Considering this load characteristic then the vibrational effects on the scanning manipulator will be reduced, when this carry out scanning tasks over the terrain.

Quadruped robot has the advantages of high speed, strong obstacle capability, high flexibility and so on. It has become a hot research field of robotics. The research of quadruped robot in our country is in a state of rapid development. The robot of large-scale high-performance and high stability is the goal of our research. But the design and research of small quadruped robots is also indispensable. At first we use SolidWorks to design the frame structure of the quadruped robot. Then we use simulation software ADAMS to achieve the quadruped robot gait of Trot. So we further determine the feasibility of the design.

Planning a robust free gait is an important thing to improve the terrain adaptability of quadruped robot. By using a robust free gait, quadruped robot can walk through the rough terrains with high complexity. The foothold search method is presented in this paper to search the optimum foothold for the swing foot, and the stability of the robot can be improved by using the proposed foothold search method. Furthermore, in order to ensure that quadruped robot can swing a foot forward with enough stability margin, just before the swing foot lift-off the ground, a four feet phase is added into the walking process of the robot to complete a body sway movement to enlarge the stability margin, and the Cubic Bezier curves are used to construct the COG trajectory. With the proposed free gait, the robot can walk through the rough terrains automatically and successfully. The correctness and effectiveness of the proposed method is verified via simulation.

This paper proposes to plan the bipedal robot sprinting using virtual forces control method owing to the toe joint between the support leg and ground is under-actuated. However, under the virtual forces control, the internal motion of the support leg becomes uncertain and the works of the joint torques may pit against each other during the tiptoe support phase in sprinting due to the added redundant toe joint. To solve this problem, a driving coordination between the knee and ankle joints method that balances the instantaneous works of the two joints by an angular velocity follower for the ankle joint is proposed. Simulations experiments verify the effectiveness of the proposed method.

This paper presents a novel foot placement control algorithm for adaptive bipedal walking. In this method, the torso attitude and height are stabilized by synergic patterns so that the forward velocity and its change have a stable and nearly linear relation with the foot placement. Hence, our proposed online linear regression analysis well represents the local linear models by estimating continuously from measured data. Based on this estimation, an appropriate foot placement can be determined to control the forward velocity. Our simulation study successfully demonstrates the natural gait with accurate tracking of walking velocity, and the robustness of walking over uneven terrain.

This paper presents a hybrid position/force control algorithm to perform motion control of the quadruped robot based on a simplified planar model in sagittal plane which can describe the main dynamical characters of trotting gait. In order to realize the approximation of body motion decoupling, the vertical position and pitch of the body are controlled by using position servo and the horizontal position control is performed by using force control. And several constraints are considered to maintain the balance of the body during motion control, so as to increase the robustness of the robot while large interference are exerted. Simulation is carried out to validate the effectiveness of the proposed algorithm in Matlab/SimMechanics.

There are lots of demands for the amphibious travelling style of mobile robots. This paper proposed a simple, elastic and useful leg for mobile robots. When moving on the ground, the leg could be deformed a curve-like one passively, forced by the ground reaction force. Moreover, the leg would recover straight, fin-like, and propel the robot in the water. The leg just needed to be fixed on the output shaft simply and its deformation could be realized without any additional mechanism. That made it could be widely applied to much more robots. To evaluate the movement performance of the leg, an experiment platform was designed and series of experiments were carried out. The results verified our assumption that the elastic leg would be useful for robot’s travelling. The data obtained in the experiments would also help optimize this leg in the future.

In this paper, a two-dimensional passive walking robot is designed and processed, and dynamics model of the robot is established to perform the numerical simulation with the parameters of the prototype. Besides, with different parameters, the numerical simulation is performed to research the impact of parameter changes on gait. Finally, we do the prototype walking experiment.

Two-wheel inverted pendulum (TWIP) robots attract public attention as environmentally-friendly products for personal transportation in recent years. This paper presents a multi DOF TWIP(MD-TWIP) robot to balance its structure as well as sliding and swinging mechanisms. To fulfill the trajectory planning task, we proposed a robust controller based on the sliding mode theory for equilibrium control. In addition, a velocity control strategy is also introduced based on the T-S fuzzy controller to achieved good results in robot motion. Results are verified in ADAMS and MATLAB platforms simultaneously. Furthermore, the proposed control strategy is implemented on a practical self-designed physical TWIP robot model.

An amphibious robot with straight compliant flipper-legs can conquer various amphibious environments. It can rotate its flipper legs and utilize the large deflection of the legs to locomote on rough terrains, and it can oscillate the straight flipper legs to propel underwater. This paper focus on the dynamics of the compliant straight flipper legs during terrestrial locomotion. Leg's motion is modeled dynamically using large deflection theory and simulated to investigate locomotion parameters including trajectory, velocity and propulsion efficiency. In order to validate the theoretical model of the locomotion dynamics, a single-leg experimental platform is set up to explore the locomotion performance of the flipper legs with various structural and kinematic parameters. The trajectories of the rotating axle of the leg during locomotion in simulation and experiment coincide approximately. The dynamical analysis in this paper for terrestrial locomotion facilitates the implementation of amphibious robots with compliant flipper legs.

Amphibious robots are attracting more and more attentions from researchers worldwide for their broad foreground in resource exploration, disaster rescue, and reconnaissance. AmphiHex-I with transformable flipper-leg composite propulsion mechanisms can adapt various terrestrial and water environments. In this paper, we explored the locomotion performance of AmphiHex-I on various terrains by dynamical simulation. The influence of the stiffness of the flipper legs was investigated comprehensively. The simulation result indicates that the propulsion efficiency of the flexible flipper leg is higher than that of the rigid curved leg when passing through stairs, the variant of center of gravity (CG) keeps lower when the robot adopt a flexible flipper leg on flat terrain, the average velocity of the robot with flexible flipper legs is higher when the initial phase difference is large. These results verify that the transformable flexible flipper-leg mechanisms enable AmphiHex-I to perform better in passing through complex terrestrial environments.

Materials evaluation to establish component service lifetime in high radiation environments is often by means of irradiation. Achieving an accelerated lifetime “dose” of radiation drives better materials selection & system performance. The software control provides the ability to irradiate any area or single points of a sample using a proton beam from a Cyclotron. Samples cooled in a thermal chamber are moved at a constant velocity along a path or static point through the beam. The beam orientation is transverse to the sample movement, intersecting exactly with specific points on the sample. Accurate timing and positioning allow a controlled dose of radiation to the sample. Secondary dose monitoring is by means of measuring beam charge in a graphite Faraday cup though out the irradiation run. Confirming the level of induced radioactivity to the sample is by measuring the accumulated radioactive nuclei decay or half-life of a Ni foil mounted adjacent to the sample inside the thermal chamber. This scanning control of the beam delivers a highly uniform dose of radiation within a specified tolerance. Additionally, scanning path profiles can be adjusted prior the irradiation of samples. The radiation dosing software control of the Pre-configured XY-axis Cartesian Robot System is described in detail in this paper.

Hand-eye calibration to determine the relative pose between the robot and camera is a crucial problem for many tasks where the robot has to physically interact with the environment. This paper introduces an autonomous method for the calibration of eye-to-hand system, which is aimed at making the hand-eye calibration easy to operate in the industrial application. In this method, motion planning based on imaging model of camera is executed to make sure the calibration object mounted on robot end-effector in the range of camera vision. The poses of calibration object and robot end-effector are gathered in real time to calibrate and update the hand-eye relation until the result is convergent. Experiments have been conducted and experimental results show the effectiveness and accuracy of the proposed method.

Quadcopter UAVs have several complex conditions including nonlinearity, verily coupled, and working area that tend to be subject to disturbances and uncertainties. Sliding Mode Control (SMC), which is one of well-known robust control approaches, is believed to be able in dealing with such issues. However, chattering phenomenon in SMC is still an issue that may lead to performance degradation. Interval Type-2 Fuzzy Logic Control which has the capability to accommodate uncertainties of the system, will play a vital role in eliminating this phenomenon. A combining of the two control methods is investigated in this paper, and the results obtained demonstrate good performance of the approach in dealing with nonlinearity, disturbance, and uncertainty as well as eliminating the chattering phenomenon of SMC.

Reinforcement Learning (RL) is commonly used in learning new skills or adapting new situations for humanoid robots as task or environment changes. However, new skills acquisition through RL usually starts from a state of tabula rasa, which is a time-exhausted procedure and can not be suitable for real physical robot, especially when task is a complicated one. Thus, with the emphasis on skills learning efficiency, utilizing past knowledge or experience instead of from a tabula rasa becomes a hot topic. Knowledge transfer learning with RL has exhibited low sample complexity and then been heavily focused. In this research, underling the knowledge transfer learning on autonomous robot, the problem of autonomously modeling inter-task mapping between source task and target task is addressed, where a three-way Restricted Boltzmann Machine (RBM) is employed. To further decrease the sample complexity of transfer leaning on humanoid robots, biased sampling technique is proposed instead of random sampling. To evaluate the performance of the contribution of this research, experiments are performed on a physical robot, PKU-HR5.1, with task domain of bipedal walking on both flat and slope surfaces. Experimental results demonstrate the effectiveness and efficiency of the proposed approach.

Accurate and real-time gait phase detection is the fundamental question of intelligent prosthesis control. A good phase detection method can make the prosthesis control better and easier . Here we propose a real-time detecting method for gait phase and build a phase detection system for prosthesis. The system consists of 4 foot force sensors, two attitude sensors and an interaction interface program. Experiments were performed on three subjects with different walking velocities and different stride length. Gait phase could be detected correctly for all the subjects. The experimental results show that the proposed method is effective and accurate.

For the intelligent control of lower limb prostheses, kinematic reference trajectories are required. Due to individual difference and various preferred reference walking-speed, a general reference knee angle trajectory is difficult to adapt different people and their different walking speed. To solve this problem, gait kinematic information of thirty healthy subjects was collected at their preferred speeds and the relationship between kinematic models of knee joint and speeds was analyzed. Each knee angle trajectory was modeled based on Fourier functions, and the parameters of models were proved to be related to walking speed. Finally, kinematic models for knee angle trajectories were reconstructed based on regression equations. The reconstructed trajectories not only can adapt to the various walking-speeds, but also adapt to different people. The result of analysis showed that the reconstructed trajectories matched the measured motion trajectories well.

Such modern social issues as demographic ageing and relevance of living conditions enhancement, along with mobility maintenance of different population groups, makes powered orthosis and exoskeleton the core research and application issue. Being widely used, they play a key role for medication purposes, such as rehabilitation of patients suffering from stroke, spinal cord injuries or lower limb surgeries…

Simulation technique of studying adaptation motion of mobile wall climbing robot with sliding seal and two wheels is suggested. Such parameters as forces, pressure, velocities of contact device have been taken into account. The results of simulation present dynamic range of the robot motion with restriction on parameters. It is shown the limitation of characteristics of robot adaptive motion over complex surfaces. The obtaining results were used for the robot's prototype design.

This paper presents a method for gait optimization of modular robotic locomotion. A physics-based simulation system for development and optimization of locomotion patterns is utilized. A graphical user interface (GUI) allows for interaction with the simulation at runtime and to supervise it. Long-term optimizations can be performed as background processes without GUI. Great benefit arises from the possibility to optimize even not well understood and complex control parameters whose correlations with the behaviour of the associated control algorithm and other parameters are not clear. The successful application of the system for control parameter optimization is shown in two different experiments. Each of these experiments is performed with two different optimization algorithms. The advantages and disadvantages of genetic algorithms versus classical reinforcement learning is discussed in the end.

LfD(Learning from Demonstration) has the advantage of requiring no expert knowledge about the robot itself, which make large-scale application of robots possible. In this paper, we present a novel approach based on combination of affine deformation and DMP(dynamic movement primitive) to deal with two fundamental problems in LfD: data gathering and policy deriving. In the proposed approach, demonstrated motion data are gathered from optical-based motion capture system and DMP is used to sketch feature and derive policy. Combined with affine deformation, joint trajectory derived from the control policy can be refined so that the manipulator's physical constraints satisfied, the end point accuracy preserved and the execution time optimized. We verify the feasibility of our approach by reproducing a series of different motions with various trajectory profiles on a humanoid robot's arm basing on limited human demonstrations.

Climbing robot, flying robot or human wearable devices usually execute daily tasks in a pre-defined workspace sharing with humans. Long-term operation for these robots posts three challenge: 3D pose estimation, cost limitation and unexpected low dynamics. To address these challenges, we propose a solution for performing multi-session SLAM using a RGBD sensor. The main model is a multi-session pose graph, which evolves over the multiple visits of the workspace. When the robot explores the new areas, its poses will be added to the graph. The poses in the graph will be pruned if their corresponding 3D point scans are out of date. Thus the scans corresponding to the poses kept in the current graph will always give a map of the latest environment. To detect the changes of the environment, an out-of-dated scans identification module is proposed. Pruning of the poses also decreases the computational burden in graph optimization. Experimental results using real world data acquired by a Kinect sensor show that the proposed framework is able to manage the map in date for low dynamic environments with a reduction in complexity and an acceptable error level compared to the method saving all poses.

This paper presents a strategy for sound source localisation using an embedded asynchronous microphone array for robotic target tracking application. Conventional microphone array technologies require a multi-channel A/D converter for inter-microphone synchronization making the technology relatively expensive. In our method, a synchronization free embedded asynchronous microphone array has released this requirement. The microphone array needs self calibration using graph-based SLAM method, which estimates starting time offset and clock difference/drift rate of each microphone channel using Gauss-Newton least square optimization. The proposed method is suitable for target tracking applications.

The covered area detection method, using the brightness change in an image and parallax correction, does not depend on a common field of vision. Camera images become important information when a rescue robot is operate by remote control, but mounting cameras is difficult on a rescue robot crawler that must get into cracks in rubble. We propose attaching cameras behind the crawler shoe. The biggest problem then, however, is that the shoe obstructs large parts of the camera image. To avoid this, we developed real-time image processing that complements the obstructed area through the use of two cameras. We then performed evaluation experiments to confirm the effectiveness of the proposed technique. The complement method can solve in a case of one parallax problem. However, several objects in a acquiring image occurs several parallaxes between objects. In this paper, we report a trial to solve the parallaxes problem by using a user interface software.

Many disasters occurred in coal mines, but the explosive gas, less oxygen and hot air stopped the rescuers from doing their job. An omnitread serpentine robot –HITSR-I was developed to work in these areas. The purpose of this paper is to introduce the mechanism and anti-explosion design of the robot. With all these unique designs, the serpentine robot can search in hazardous areas. And the experiments have been carried out under coal mines.

In this paper we present a navigation framework for field deployment for the hydraulically actuated quadruped, HyQ. Our framework uses a lidar sensor to perceive obstacles and free space around the robot and plans a path through traversable areas to given goals. A path following procedure generates velocity and angular rate commands that are sent to a locomotion controller that produces the trotting gait pattern, following the desired path to the goal.

A strategy for finding the most reliable path is proposed to deal with target search problem in this paper. When a robot chases a target in Semi-structured environments with uncertainty and hazards, different roads may lead to disparate risks. A kind of reliability based topological map is designed, in which uncertainties such as road condition and threats are considered. And then, a motion planning method is presented to minimize the expected-risk in carrying out a target search task. Further simulation and physical-world experiments are presented to demonstrate the feasibility and validity of this approach.

This paper proposes a control strategy based on velocity resolution for velocity and orientation tracking of a trotting quadruped robot. In trotting gait, the pair of diagonal stance legs and body of robot can be approximately modeled as a virtual planar-sevenlink mechanism and a SLIP model by projecting them into two orthogonal planes. To reduce tracking error of velocity, the foot placements are computed based on SLIP model. As there is misalignment between the centroid of body and the geometric center of body, the core of the trotting quadruped robot is switching control in the option of this paper. Based on Lyapunov theory, definition of stable region and sufficient condition of convergence of trotting quadruped robots are proposed. Finally, an experiment is performed to validate the effectiveness of proposed strategy.

In order to alleviate the counteractive effect on the body of humanoid biped robot while the humanoid manipulator operates under rapid continuously reaction environment (e.g. baseball playing, ping-pong playing, etc.), a trajectory planning method with minimum-acceleration based on differential evolution (DE) is presented in this paper for humanoid manipulator with seven degrees of freedom (DOFs). In this method, quadrinomial polynomial is used to describe the segments of the operation process and return process of the manipulator movement at joint space. Through optimizing the angular acceleration of the trajectory at the target point moment using DE, minimum-acceleration trajectory with continuous profile property is obtained, which can reduce the disturbance against the humanoid biped robot to a certain extent. Simulation experiment results show the effectiveness of this method for the trajectory planning problem studied.

The paper presents an analysis modeling and trajectory planning for a car-like climbing robot used for stripping coatings from the outer surface of underwater pipes used in oil platform offshore. The simple modeling method we proposed is based on the mapping relationship between modeling in plane and in cylinder, and using the method we designed trajectories for the robot to travel all over the outer surface of the pipes completely and efficiently. Simulation and experimental tests were carried out in order to verify the satisfaction of the analysis modeling and to validate performance of the control system. Although the method worked well in most cases, more factors and conditions need to be taken into consideration in the control in the future works.

Successful deep-sea exploration (up to 6000m) performed by autonomous underwater vehicles (AUVs) depends heavily on accurate AUV localization. In the absence of GPS underwater, this requires acoustic communication, localization and geo-referencing underwater participants, e.g. by a GPS-positioned surface vehicle, which is necessary to periodically correct drift errors from deadreckoning. Within the SMIS project this is supposed to be achieved by a team of AUVs assisted by an unmanned surface vehicle (USV) and a sea-bed-station. In order to derive constraints for the team behaviour, the quality and stability of the acoustic localization needs to be examined thoroughly in advance. This delivers various constraints for later at-sea operation, e.g. the maximum distance between AUV and USV and the maximum horizontal distance to the USV for a given AUV depth. Additionally, heuristic estimates for reproducibility of long-range position measurements can be derived which indicate imminent errors based on acoustic localization only. This paper reports associated tests and their results using USBL modems in deep-sea trials performed in the Atlantic ocean.

Pectoral fin motion discipline of the cownose ray as a nature prototype is analyzed. According to the analysis results, a bionic pectoral fin with multiple fin rays and one revolute wingtip is designed. The angle of attack distribution principle of the bionic pectoral fin is also analyzed. The analysis results show that the angle of attack of pectoral fin reaches its maximum at the wingtip. In order to improve the deformation ability of the tip of rigid fin ray, a novel revolute joint is designed and applied to the bionic pectoral fin. The simulation results show that the maximum angle of attack of the revolute wingtip can reach about 35°.The relationship between the pitching angle of fin tip and the average thrust/lift generated by pectoral fins is analyzed. In the thrust/lift experiments, with the help of the revolute joint, the maximal average thrust of 3.3N is achieved and the maximal absolute value of average lift is less than 2N. In the swimming experiments, a maximum forward speed of 0.7BL/s (body length/s) is reached. Under the same motion parameters, the maximum speed decreases to 0.35BL/s when the revolute joint is locked.

This paper presents an innovative design concept for a biomimetic dolphin-like underwater glider. As an excellent combination, it offers the advantages of both robotic dolphins and underwater gliders to realize high-maneuverability, high-speed and long-distance motions. As the first step, a skilled and simple dolphin-like prototype with only gliding capability is developed. The hydrodynamic analysis in the glider using Computational Fluid Dynamics (CFD) method is executed to explore the key hydrodynamic coefficients of dolphin-like glider including lift, drag and pitching moment, and also to analyze the dynamic and static pressure distribution. Finally, experimental results have shown that the dolphin-like glider could successfully glide depending on the pitching torques only from buoyancy-driven system and controllable fins without traditional internal movable masses.

In this paper, we present the mechanical design of a novel robotic fish capable of fast swimming and yet with less joints. Prior to designing, the principle and goals for design are analyzed, in which the size, central pattern generator (CPG) model, and oscillatory frequency of robotic fish play an important role. Then, the mechanical structure of the two-joint robotic fish is designed. The first joint of the robotic fish is driven by a gear motor. The gear motor is linked to an eccentric wheel which rotates in a free slide. The second joint is driven by a connecting rod to make the two joints form an angle between them, allowing free direction adjustment and flexibility. After that, the mechatronic design and CPG-based control are described. Underwater tests are performed on the robotic fish, validating the effectiveness of the proposed design scheme. Particularly, the robotic fish reached a maximum speed of 0.7 body lengths per second, which is expected to be much faster after improvement.

Hydraulic-driven Bionic Undulating Robot (HBUR) is regarded as one of fish-inspired underwater robots, whose bionic fin rays are actuated by a specially designed hydraulic system, and aims to promote the flexible performance of bionic motion and hydrodynamics. To understand in-depth the undulating mechanism of the bionic hydraulic system, so as to develop a more advanced bionic structure and valve system, this paper concentrates on model developing and morphology analyzing of undulating motions on HBUR. First of all, mathematical models for HBUR are setup with considerations on mechanism of bionic valve, as well as on bionic structure of hydraulic swing actuator (HSA). With some experiments, these analytic models were verified. Then, undulating morphologies and its hydrodynamics were investigated and discussed, to evaluate the actuating performances of the bionic hydraulic system. Results show that the specially designed valve could actuate HSA swinging and HBUR undulating as expected, under a much more simplified mechanical structure compared to the servovalve approach. Characteristics of morphologies and hydrodynamics indicate that HBUR will be very promising for implementing a more progressive undulating robot underwater.

The following sections is included:

• AUTHOR INDEX