基于矢量场的无人机动力学规划算法.

Aiming at the problems that trajectory generated by traditional hierarchical motion planning was limited to a given homology class and rapidly-exploring random tree had a lower efficiency of sampling, this paper proposed a kinodynamic planning algorithm based on vector field guided sampling. Firstly, this algorithm constructed a cone by introducing vector filed which defined by the gradient of potential field function, and utilized the cone to constraint the sampling of RRT * . Then, solving the optimal control problem could generate motion primitives to handle the two-point boundary value problem,it derived an explicit solution for the optimal duration-time of the motion primitives so as to optimally connect any pair of states. Finally, this paper drove a quadrotor in MATLAB simulation environment to verify the feasibility of the generated trajectory. Systematic comparisons with the state-of-art kinodynamic planning algorithm demonstrate that the proposed algorithm explores more states with shorter running time and the generated trajectory has smaller arrival-time and control effort under the same number of iterations. [ABSTRACT FROM AUTHOR]