First, the FM* algorithm appears to be the best static trajectory planning algorithm (better than A*) both in terms of computation time and smoothness of the trajectory solutions. Second, dynamic trajectory planning algorithms are faster than static planners after a limited amount of time. Third, the DFM algorithm is slower than the D* Lite algorithm but it allows the curvature of the trajectories to be controllable.
Contrary to D* Lite, DFM algorithm does not include any heuristic to speed up the exploration process. One interesting direction for further research would be to develop what could be called the DFM* algorithm. A novel DFM* algorithm would combine the accuracy of the DFM algorithm with the exploration efficiency of the FM* algorithm.
Was this article helpful?