On July 10^th, 2023 at 11.50 am Andrea Ticozzi, DEIB PHD Student, will give a seminar on "Global Overtake Planning for Autonomous Vehicle Racing on Ovals" in DEIB Conference Room.

Head-to-head autonomous vehicle racing is quickly becoming a fundamental proving ground to test autonomous driving technologies at the limits of their capabilities. From control to state estimation, passing through perception and trajectory planning, all the aspects of the driving algorithms are stressed in a challenging and competitive environment. With this presentation, I will introduce a novel global planning algorithm to effectively plan overtaking maneuvers when racing at high speed on oval tracks. By making use of a track-based occupancy grid framework, the algorithm first learns the nominal path followed by the opponent. Then, it uses this information to plan a globally optimal attack plan through a graph-based search of possible overtaking paths along the track, exploiting offline-computed information on the track three-dimensional features. I will conclude the presentation with some results obtained on a full-scale autonomous racing vehicle in the context of the Indy Autonomous Challenge; such results show the algorithm overtaking capabilities against several autonomous opponent drivers, up to a top-speed overtake at 275 km/h.