Guided Motion Planner
Duong Le / April 2020 (143 Words, 1 Minutes)
This project focuses on motion-planning problems for high-dimensional mobile robots with nonlinear dynamics operating incomplex environments. It is motivated by a recent framework that combines sampling-based motion planning in the state spacewith discrete search over a workspace decomposition.
The using probabilistic roadmap abstractions in combination with sampling-based motion planning to effectively plan collision-free and dynamically-feasible robot motions to reach a given goal. Roadmap abstractions are constructed over a low-dimensional configuration space obtained by considering relaxed and simplified representations of the robot model and its underlying motions. By capturing the connectivity of the corresponding free configuration space, roadmap abstractions provide the approach with promising suggestions of how to effectively expand the tree-based search in the full state space of the robot.
The approach is geared toward high-dimensional motion planning problems for mobile robots with nonlinear dynamics. Experimental results showed significant computational speedups when using roadmap abstractions instead of workspace decompositions as well as when comparing the proposed approach to other successful motion planners. The combination of roadmap abstractions in configuration spaces with tree-based exploration of the state space opens up new venues for research. In fact, roadmap planners have been extensively improved over the years to handle increasingly complex motion-planning problems in configuration spaces. By taking advantage of this extensive body of literature, we plan to improve the sampling and connection strategies used in the roadmap construction to more effectively capture the connectivity especially inside narrow passages.
Simulation with physic-based vehicle
Simulation with high-dimensional robot (snake-like robot)
Simulation with high-dimensional robot (blimp-like robot)