No abstract

Humans are very skilled at learning new control tasks, and in particular, the use of novel tools. In this article we propose a paradigm that utilizes this sensorimotor learning capacity to obtain robot behaviors, which would otherwise require manual programming by experts. The concept is to consider the target robot platform as a tool to be controlled intuitively by a human. The human is therefore provided with an interface designed to make the control of the robot intuitive, and learns to perform a given task using the robot. This is akin to the stage where a beginner learns to drive a car. After human learning, the skilled control of the robot is used to build an autonomous controller so that the robot can perform the task without human guidance. We demonstrate the feasibility of this proposal for humanoid robot skill synthesis by showing how a statically stable reaching skill can be obtained by means of this framework. In addition, we analyze the feedback interface component of this paradigm by examining a dynamics task, in which a human learns to use the motion of the body to control the posture of an inverted pendulum that approximates a humanoid robot, so that it stays upright.

This paper proposes an effective framework of human-humanoid robot physical interaction. Its key component is a new control technique for full-body balancing in the presence of external forces, which is presented and then validated empirically. We have adopted an integrated system approach to develop humanoid robots. Herein, we describe the importance of replicating human-like capabilities and responses during human-robot interaction in this context. Our balancing controller provides gravity compensation, making the robot passive and thereby facilitating safe physical interactions. The method operates by setting an appropriate ground reaction force and transforming these forces into full-body joint torques. It handles an arbitrary number of force interaction points on the robot. It does not require force measurement at interested contact points. It requires neither inverse kinematics nor inverse dynamics. It can adapt to uneven ground surfaces. It operates as a force control process, and can therefore, accommodate simultaneous control processes using force-, velocity-, or position-based control. Forces are distributed over supporting contact points in an optimal manner. Joint redundancy is resolved by damping injection in the context of passivity. We present various force interaction experiments using our full-sized bipedal humanoid platform, including compliant balance, even when affected by unknown external forces, which demonstrates the effectiveness of the method.

With the ultimate goal of producing natural-looking movements in humanoid robots and virtual humans, we examined the visual perception of movements generated by different models of movement generation. The models of movement generation included 14 synthetic motion generation algorithms based on theories of human motor production. In addition, we obtained motion from recordings of actual human movement. The resulting movements were applied to both a humanoid robot and a computer graphics virtual human. The computational efficiency of the motion production algorithms is described. In Experiment 1, we examined observers' judgments of the naturalness of a movement. Results showed that, for the humanoid robot, low ratings of naturalness were obtained for rapid movement. In addition, it was found that some movements that appeared to have unremarkable naturalness ratings were anomalous examples of the desired movement. In Experiment 2, we used naturalness ratings to study the influence of movement speed on the humanoid robot. Results indicated that the decrease in naturalness was due to motion artifacts at the ends of the movement. In Experiment 3, we returned to the issue of anomalous movements by obtaining ratings of similarity between pairs of movements, and analyzing these with multi-dimensional scaling to obtain a psychological space representation of the set of movements. Results showed that the presumed anomalous movements were indeed distinctive from the other movements, suggesting that the naturalness judgments did not completely indicate the perception of movement. We discuss these results in the context of what they suggest for the relative effectiveness of the different generation algorithms at producing natural movement, and their relative computational efficiency, as well as in terms of the effectiveness of different psychological techniques for the assessment of humanoid movement.

No abstract