Hybrid Mobile Robot Controller for Reactive Navigation in Unknown Environment

Reactive navigation task in an unknown environment is a generalized formulation of the navigation task that assumes no information is available on the robot environment, which includes the possibility of recalculating the currently followed path in real time due to approaching previously unknown obstacles and not being able to perform according to pre-planned path. This work presents several algorithms - PRM-D* with resampling strategy, simple motion controller, and convex hull-based heuristic, which create a robust navigation system. PRM-D* hybrid with resampling strategy included allows for fast path replanning in real-time. Motion controller processes LiDAR data in a simplified manner, which allows for modifying trajectory without spending a big amount of processing power. The convex hull algorithm was used as a heuristic to determine feasibility of the path and decide whether the robot should try to approach the current waypoint, or proceed with navigation toward the next point.