FBGA: Fast and Flexible Velocity Planning for Time-Optimal Maneuvers

Mattia Piazza, Mattia Piccinini, Sebastiano Taddei, Francesco Biral, Enrico Bertolazzi

Keywords: velocity planning, real-time optimization, time-optimal maneuvers, forward-backward method

Abstract:

Time-optimal velocity planning under generic
acceleration constraints is critical for autonomous
racing. Existing methods face a trade-off: optimal
control handles complex constraints accurately but is
computationally expensive, while fast semi-analytical
methods are limited to conservative box constraints. We
propose FBGA, a Forward-Backward algorithm with
Generic Acceleration constraints, discretizing the path and
performing forward-backward passes to maximize velocity
profiles. Tested on five racetracks with racing cars and
motorcycles, FBGA matches optimal control accuracy (within
0.11%-0.36%) while achieving 2-3 orders of magnitude
speed-up, making it suitable for real-time multi-query
trajectory planning. Open-source C++ implementation:
https://github.com/DRIVEWISE/FBGA.

Submission Number: 14