Incremental Adaptive Probabilistic Roadmaps for Mobile Robot Navigation under Uncertain Condition

摘要: As the application domains of sampling-based motion planning grow, more complicated problems arise that challenge functionality these planners. One main challenges is weak performance when reacting to uncertainty in robot motion, obstacles, and sensing. In this paper, a multi-query planner presented based on optimal probabilistic roadmaps algorithm employs hybrid sample classification self-adjustment strategy handle diverse types uncertainty. The proposed method starts by storing collision-free generated samples matrix-grid structure. Using resulted grid structure makes it computationally cheap search find specific region. soon as senses an obstacle during execution initial plan, occupied cells are detected, relevant selected, in-collision vertices removed within vision range robot. Furthermore, second layer nodes connected current direct neighbors checked against collision which gives time react before getting too close obstacle. simulation results with show significant improvement comparing similar algorithms terms failure rate, processing minimum distance from obstacles. was also successfully implemented TurtleBot two different scenarios