International Journal of Humanoid Robotics

Generally, robots are initially designed via computer-aided software to obtain their parameters. However, the given parameters are not entirely true due to their unmodeled parts; therefore, system identification is needed, and physical conditions are crucial to guarantee feasible solutions. This study uses a quadratic programming regression model accompanied by physical consistency constraints and designs specific exciting motions for humanoid robots. The proposed constraints are designed based on the geometric approximation of link objects, the physically reasonable mass and inertia and the total mass of the robot being correct. The proposed exciting motions include the general walking motion and the single-leg support motion, which enable a more flexible and stable way to cause excitation in the floating-base system. The identified parameters are evaluated on the humanoid robot NINO. Furthermore, the error between the feedback information of the zero moment point and the command information of the center of mass are used for evaluating the identified dynamic parameters. According to the experiments with the proposed exciting motions, the identified parameters are found to be obviously better than the original computer-aided design parameters, especially in the $x$ direction.

The dynamic force distribution is significant in multi-fingered hand operation. It is essential to choose the appropriate contact force between fingers and object to ensure both stability and power consumption requirements corresponding to the specific task. We describe nonlinear contact force distribution problems as smooth manifold optimization problems corresponding to a linearly constrained positive definite matrix. A quadratic index gradient flow approach is presented to optimize the contact force based on a low-dimensional description matrix. And, a task-oriented contact stability criterion is proposed to evaluate contact stability. A simple selection method of the weight parameters of the objective function is provided for the specified contact stability criterion. The proposed algorithm has better force optimization results and fewer iterations than the traditional gradient flow optimization algorithm under the same stability condition. Numerical examples demonstrate the correctness and effectiveness of the algorithm. The proposed method of dynamic force distribution can be used for the real-time application of multi-fingered grasping and manipulation.

We consider the whole-body humanoid gait control model via reinforcement learning, where an initial predefined parametric gait policy is optimized by employing the augmented random search algorithm realizing most of the experiments in simulation and verifying the results onto the real robot. In our proposal, we take into account the reality gap problem, also known as Sim to real problem, for the development of the experiments and the policy guaranteeing functional results in the real robot employing a simplified custom simulator. We show that our proposed learning algorithm allows the humanoid robot MAX-E2 to walk efficiently in virtual and real environments.