Jerk-Limited Time-Optimal Speed Planning for Arbitrary Paths

Both path and speed planning play key roles in automated driving, specially when a certain comfort level is required. This paper proposes a human-like speed planning method for already defined paths while minimizing travel time. The proposed method is able to compute a time-optimal speed profile that meet the given constrains with regard to speed, acceleration and jerk. For this purpose, an initial acceleration-limited approach is introduced. This algorithm serves as a starting point for the subsequent jerk-limited speed planning. Moreover, fallback strategies are included to manage critical driving situations where initial or final conditions cannot be met. The proposed approach has been tested and validated in an experimental platform through extensive trials in real environments. Its performance has been evaluated both in terms of quality of the computed speed profiles and with respect the required computing time.