Path planning is an active and essential research field for many applications with autonomous mobile robotics. However, popular technologies have limitations in situations where robots require onboard computing to work independently. To this end, this paper proposes a bidirectional advanced batch information tree (BABIT*), which is an asymptotically optimal algorithm path planner enhanced from Batch Informed Trees (BIT*). It uses an edge queue sorted by inflated potential path cost to guide the search of implicit random geometry graph (RGG) to generate explicit solutions while minimizing height calculation tasks such as collision checking. BABIT* promotes the exploration of the entire state space by adopting a more reasonable sampling strategy to achieve a more uniform and decentralized approximation of problems, and ensures a faster discovery of solutions by using symmetric bidirectional search for the state space from both directions. The experimental results show that BABIT* outperforms existing single-query, sampling based planners on the tested problems.
|