Sensor-based motion planning with kinematic and dynamic constraints.

摘要

This Ph.D. thesis addresses theoretical, algorithmic, and implementation issues of the role of system kinematics and dynamics in sensor-based motion planning. The emphasis is on real-time planning, on fast operation in a complex environment with unknown obstacles of arbitrary shapes, and on the efficiency of generated motion.; Today's sensor-based planning systems largely cannot handle dynamics and ignore the influence of various (e.g. nonholonomic) constraints on the robot motion planning. Consequently, they can be used only in those few applications where dynamics or such constraints are either absent or unimportant--which is rather rare in real-world tasks. The goal of this work is to remove these limitations.; In particular, three problems of two-dimensional sensor-based motion planning are solved: (1) an algorithmic approach is developed for handling robot dynamics in the context of sensor-driven holonomic systems; (2) a real-time solution is proposed of the shortest path problem for nonholonomic systems operating in an unlimited workspace; (3) a new problem in nonholonomic motion planning is introduced--curvature-constrained motion planning within a limited workspace; a methodology for handling this problem is developed, and constructive geometric and analytic solutions are produced.; The results of this work are expected to fill the existing gap between the availability of geometric approaches to sensor-based motion planning and desire for using these approaches in real-world applications. Theoretical results developed in the thesis can also be viewed as an attempt to provide a critical connection between control theory and dynamic systems analysis, on the one hand, and computational geometry, complexity theory, and artificial intelligence tools, on the other hand. On the application side, the developed methodology can be used in fully autonomous moving machinery (such as mobile robots and arm manipulators) as well as in human-machine systems (e.g. automobile control on smart highways and air traffic management systems).