Nonlinear Control and Identification
Final rept. 15 Aug 1999-31 Jul 2002
MASSACHUSETTS INST OF TECH CAMBRIDGE DEPT OF ELECTRICAL ENGINEERING AND COMPUTER SCIENCE
Pagination or Media Count:
This research is motivated by the problem of motion planning of autonomous vehicles in an uncertain environment. A possible approach to reduce the computational complexity of the motion planning problem for a nonlinear, high dimensional system, is based on a quantization of the system dynamics, in the sense that we restrict the feasible nominal system trajectories to the family of time-parametrized curves that can be obtained by the Inter-connection of appropriately defined primitives. These primitives will then constitute a maneuver library from which the nominal trajectory will be constructed. Instead of solving an optimal control problem over a high- dimensional, continuous space, we solve a mixed integer programming problem, over a much smaller space. At the core of the control architecture lies a hybrid automaton, the states of which represent feasible trajectory primitives for the vehicle. Each constituent subsystem of the hybrid controller will be the agent responsible for the maneuver execution. The task of the automaton will be the generation of complete, feasible and optimal trajectories, via the interconnection of the available primitives. Apart from the reduction in computational complexity, one of the objectives of this approach is the ability to provide a mathematical foundation for generating a provably stable hierarchical system, and for developing the tools to analyze robustness in the presence of uncertainty in the process as well as in the environment.
- Flight Control and Instrumentation
- Theoretical Mathematics
- Human Factors Engineering and Man Machine Systems