In this paper, we present a novel methodology to obtain imitative and innovative postural movements in a humanoid based on human demonstrations in a di®erent kinematic scale. We collected motion data from a group of human participants standing up from a chair. Modeling the human as an actuated 3-link kinematic chain, and by de¯ning a multiobjective reward function of zero moment point and joint torques to represent the stability and e®ort, we computed reward pro¯les for each demonstration. Since individual reward pro¯les show vari-ability across demonstrating trials, the underlying state transition probabilities were modeled using a Markov chain. Based on the argument that the reward pro¯les of the robot should show the same temporal structure of those of the human, we used di®erential evolution to compute a trajectory that ¯ts all humanoid constraints and minimizes the di®erence between the robot reward pro¯le and the predicted pro¯le if the robot imitates the human. Therefore, robotic imitation involves developing a policy that results in a temporal reward structure, matching that of a group of human demonstrators across an array of demonstrations. Skill innovation was achieved by optimizing a signed reward error after imitation was achieved. Experimental results using the humanoid HOAP-3 are shown.

Humans are known to manage postural movements in a very elegant manner. In the task of standing up from a chair, a humanoid robot can benefit from the variability of human demonstrations. In this paper we propose a novel method for humanoid robots to imitate a dynamic postural movement demonstrated by humans. Since the kinematics of human participants and the humanoid robot used in this experiment are different, we solve the correspondence problem by making comparisons in a common reward space defined by a multimodal reward function composed of balance and effort terms. We fitted a fully actuated triple inverted pendulum to model both human and robot. We used Differential Evolution to find the optimal articular trajectory that minimizes the Kullback-Leibler difference between the human's and robot's reward profile subject to constraints.

No abstract

There is neuroscientific evidence to suggest that imitation between humans is goal-directed. Therefore, when performing multiple tasks, we internally define an unknown optimal policy to satisfy multiple goals. This work presents a method to transfer a complex behavior composed by a sequence of multiple tasks from a human demonstrator to a humanoid robot. We defined a multi-objective reward function as a measurement of the goal optimality for both human and robot, which is defined in each subtask of the global behavior. We optimize a sequential policy to generate whole-body movements for the robot that produces a reward profile which is compared and matched with the human reward profile, producing an imitative behavior. Furthermore, we can search in the proximity of the solution space to improve the reward profile and innovate a new solution, which is more beneficial for the humanoid. Experiments were carried out in a real humanoid robot.

There is an open discussion between those who defend mass distributed models for humanoid robots and those in favor of simple concentrated models. Even though each of them has its advantages and disadvantages, little research has been conducted analyzing the control performance due to the mismatch between the model and the real robot, and how the simplifications affect the controller's output. In this paper we address this problem by combining a reduced model of the humanoid robot, which has an easier mathematical formulation and implementation, with a fractional order controller, which is robust to changes in the model parameters. This controller is a generalization of the well-known PID structure obtained from the application of Fractional Calculus to control, as will be discussed in the paper. This control strategy guarantees the robustness of the system, minimizing the effects from the assumption that the robot has a simple mass distribution. The humanoid robot is modeled and identified as a triple inverted pendulum and, using a gain scheduling strategy, the performances of a classical PID controller and a fractional order PID controller are compared, tuning the controller parameters with a genetic algorithm.