Intelligent control and learning of robots interacting with environments

摘要

The objective of this research is to investigate intelligent learning and control for robots interacting with environments. In the first part of this thesis, impedance and trajectory adaptation are investigated independently. Impedance adaptation is developed using a cost function or a reward function to describe the interaction performance, and impedance parameters to be adjusted to maximize the reward function. For the proposed reference adaptation, the cost function is minimized using trajectory parametrization and iterative learning. Besides physical interaction, when the robot is navigating in human environment, social rules and constraints need also to be addressed for friendly and natural robot motion control. In the second part, a novel control scheme based on the social force model is proposed for robots navigating in human environments. To address the constrained motion problem of robots, a combined kinematic/dynamic control is proposed for robot motion control which is subject to ellipsoidal position and velocity constraints.