Motion planning for aggressive autonomous vehicle maneuvers
Aliasghar Arab, Kaiyan Yu, Jingang Yi, Dezhen Song · 2016
We present a motion planning algorithm for autonomous aggressive vehicle maneuvers. The motion planner takes advantages of the sparse stable trees (SST), the RRT* algorithm and the model predictive control (MPC) design. The use of the sparsity property helps to reduce the computational burden of the RRT* method by removing non-useful nodes in each iteration (i.e., rewiring) and therefore to quickly converge to the optimal solution. A goal-biased input is used to achieve a fast convergence. A nonlinear MPC is used to compute and find the attracting area for the generated trajectory with consideration of constraints of the system. We implement and demonstrate the motion planning algorithm on a 1/7-scale racing vehicle for autonomous aggressive maneuvers.