Final Project for the course in Autonomous and Mobile Robotics A.Y. 2020/2021.
RRT-based planners, at each iteration, generate a subpath in the configuration space, and extend the tree only if no collision occurs along it. The aim of this project is to implement a RRT-like planner that uses an optimization-based steering method which incorporates an explicit collision avoidance constraint formulated via Control Barrier Functions (CBFs), a tool that has been recently introduced for formulating safety-related constraints. In principle, this allows to guide the generation of subpaths in the free configuration space, thus reducing the number of failed extension attempts and eliminating the need of collision checks. The planner has been be tested through V-REP/CoppeliaSim simulations in different scenarios using a nonholonomic system (e.g., a unicycle) and a critical comparison with the basic RRT planner has been carried out.
The main file is v_repExtAMRProject
A detailed report of the project
The presentation we gave
- G. Giunta
- E. Nicotra
- A. Paggetti