Narrow Results

Stair climbing is still a challenging task for humanoid robots, especially in unknown environments. In this paper, we address this problem from perception to execution. Our first contribution is a real-time plane-segment estimation method using Lidar data without prior models of the staircase. We then integrate this solution with humanoid motion planning. Our second contribution is a stair-climbing motion generator where estimated plane segments are used to compute footholds and stability polygons. We evaluate our method on various staircases. We also demonstrate the feasibility of the generated trajectories in a real-life experiment with the humanoid robot HRP-4.

Unsteady locomotion and the dynamic environment are two problems that block humanoid robots to apply visual Simultaneous Localization and Mapping (SLAM) approaches. Humans are often considered as moving obstacles and targets in humanoid robots working space. Thus, in this paper, we propose a robust dense RGB-D SLAM approach for the humanoid robots working in the dynamic human environments. To deal with the dynamic human objects, a deep learning-based human detector is combined in the proposed method. After the removal of the dynamic object, we fast reconstruct the static environments through a dense RGB-D point clouds fusion framework. In addition to the humanoid robot falling problem, which usually results in visual sensing discontinuities, we propose a novel point clouds registration-based method to relocate the robot pose. Therefore, our robot can continue the self localization and mapping after the falling. Experimental results on both the public benchmarks and the real humanoid robot SLAM experiments indicated that the proposed approach outperformed state-of-the-art SLAM solutions in dynamic human environments.

Joints’ backdrivability is desired for robots that perform tasks contacting the environment, in addition to the high torque and fast response property. The electro-hydrostatic actuator (EHA) is an approach to realize force-sensitive robots. To experimentally confirm the performance of a biped robot driven by EHAs, we developed the fully electro-hydrostatically driven humanoid robot Hydra. In this paper, we evaluate the whole-body control performance realized by integrating encoders, pressure sensors, and IMU through a high-speed communication bus to the distributed whole-body control system. We report the first example of bipedal locomotion by an EHA-driven robot in both position-controlled and torque-controlled approaches. The robot could keep the balance even when the ground condition was changing impulsively and utilize its high joint backdrivability to absorb a disturbance by the null space compliance. We also report practical challenges in implementing compliant control in real hardware with limitations in parameter accuracy, torque, and response. We experimentally confirmed that the resolved viscoelasticity control (RVC), which has indirect feedback of operational space tasks by projecting the operational space feedback gain to the joint space one, was effective to tune a proper gain to stabilize the center-of-mass motion while avoiding joint-level oscillation invoked by the control bandwidth limitation. The attached multimedia file includes the video of all experiments presented in the paper.

Evolution of the human brain emerged by the pressure for communication. It is hard to believe that the pressure of evolution exempted the features of motion patterns from using them for communication. The communication skill is therefore anthropomorphic.

The author would like to call the natural function of the brain to perceive, recognize, understand and respond to the human-like motion patterns, the Anthropomorphic Biological Equipment. The natural and general humanmachine interface can be established by developing the similar function on the machine side, which we may call the Anthropomorphic Artificial Equipment. Our research started from the mathematical model of mirror neurons and continues to acquisition of semiology of human behaviors based on technologies such as unsupervised segmentation, iterative clustering, and construction of state transition network. The current research interests target to connect the behavioral semiology to the semiology of a natural language to develop a statistical system that evokes mutual association. This talk introduces the scope of our research and overviews the direction of research.

Note from Publisher: This article contains the abstract only.