Section: New Results

Motion planning in presence of highly dynamic obstacles with uncertain motion

Participants : Pierre de Beaucorps, Renaud Poncelet, Anne Verroust-Blondet, Fawzi Nashashibi.

Safe motion planning in a dynamic environment is of great importance in many robotics applications. This year, we have worked in two directions:

• The work on reachable interaction sets introduced in [37] has been extended to the case of dynamic obstacles with uncertain motions. We consider that the obstacles have stochastic motions and we use a probabilistic formulation to compute the RIS at each time step. Our approach improves existing methods in such a context (cf. Pierre de Beaucorps PhD thesis [7]).

• Focusing on autonomous vehicles, we begun to study scenarios with occluded dynamic obstacles.