Banca de QUALIFICAÇÃO: ALYSSON PAULO HOLANDA LIMA

Probabilistic foam-based trajectory planning for autonomous mobile robots in dynamic environments

Probabilistic Foam, Path Planning, Dynamic Environments.

This article presents an improved version of the Probabilistic Foam Method (PFM) aimed at planning robot paths. However, it is important to highlight that PFM, in its classical form, does not incorporate direct considerations about the presence of dynamic obstacles in the environment. In a simplified description of PFM, bubbles that propagate through free space constitute a structure called probabilistic foam, establishing an unobstructed path between an initial configuration and a final configuration. These bubbles guarantee a safe region, meeting safety requirements during maneuvers. In this approach, the free space is partially covered by overlapping convex bubbles, forming the probabilistic foam, and a search tree is constructed by randomly propagating the foam, identifying the optimal path between the specified initial and final configurations. This article proposes improvements to the PFM foam propagation strategy, aiming to adapt it to dynamic environments. The modifications aim to improve the method’s ability to deal with moving obstacles, making path planning more robust and efficient in non-static scenarios. The bubble expansion process is modified to meet spatial and temporal requirements. On the surface of a bubble, a set of daughter bubbles is expanded, with the radius determined by the difference in spatial and temporal coordinates. Thus, the article adapts PFM to trajectory planning problems for mobile robots in dynamic environments. The proposed method stands out for its ability to quickly determine ideal solutions for trajectory planning problems in dynamic environments, considering different obstacles with varying but uniform speeds. Furthermore, the algorithm provides the robot’s speed for each segment of the trajectory. The effectiveness of the algorithm is validated through functional tests in simulations, where parameters such as the minimum radius of the bubbles, the maximum speed of the robot, and the speeds and orientations of moving obstacles are defined. The simulation results demonstrate the feasibility and effectiveness of the proposed method in generating safe trajectories when the scene dynamics are known, highlighting its ability to efficiently deal with dynamic environments and providing robust trajectory planning in situations subject to changes over time.