Automated vehicle driving on large lane-free roundabouts

Automated vehicle driving on large, complex, lane-free roundabouts is a major challenge. As a striking example for this challenge, we consider the famous roundabout of Place Charles de Gaulle in Paris, featuring a width of 38
$m$
and comprising a dozen of entering/exiting radial streets. The paper proposes a complete generic methodology to control the lane-free paths of automated vehicles. The developed real-time vehicle movement control strategy relies on appropriate automated offline computation of: (a) wide overlapping movement corridors, one for each Origin-Destination (OD) movement, which delineate the admissible movement zones of corresponding OD vehicles; (b) desired vehicle orientation at each location within each OD corridor. Real-time vehicle movement within the respective corridor is effectuated by a distributed (per vehicle) nonlinear feedback control strategy, such that vehicles can move forward efficiently, accounting, when possible, for the pre-specified desired orientation, while avoiding collisions with other vehicles. Boundary controllers, developed based on linear state-feedback approaches, are used as safety filters defining upper and lower bounds for the vehicle steering angle, such that it is guaranteed that: a vehicle never violates its admissible corridor and roundabout boundaries; and never misses its exit. Microscopic simulation testing results demonstrate the pertinence and effectiveness of the suggested approach.