On infusing reachability-based safety assurance within planning frameworks for human–robot vehicle interactions

Abstract

Action anticipation, intent prediction, and proactive behavior are all desirable characteristics for autonomous driving policies in interactive scenarios. Paramount, however, is ensuring safety on the road: a key challenge in doing so is accounting for uncertainty in human driver actions without unduly impacting planner performance. This article introduces a minimally interventional safety controller operating within an autonomous vehicle control stack with the role of ensuring collision-free interaction with an externally controlled (e.g., human-driven) counterpart while respecting static obstacles such as a road boundary wall. We leverage reachability analysis to construct a real-time (100 Hz) controller that serves the dual role of (i) tracking an input trajectory from a higher-level planning algorithm using model predictive control, and (ii) assuring safety by maintaining the availability of a collision-free escape maneuver as a persistent constraint regardless of whatever future actions the other car takes. A full-scale steer-by-wire platform is used to conduct traffic-weaving experiments wherein two cars, initially side-by-side, must swap lanes in a limited amount of time and distance, emulating cars merging onto/off of a highway. We demonstrate that, with our control stack, the autonomous vehicle is able to avoid collision even when the other car defies the planner’s expectations and takes dangerous actions, either carelessly or with the intent to collide, and otherwise deviates minimally from the planned trajectory to the extent required to maintain safety.

Provably Safe and Efficient Motion Planning with Uncertain Human Dynamics

“…In the field of HAMP, Leung et al [28] developed a reachability-based controller that assures human safety by maintaining the availability of a collision-free maneuver, given a known human model. Our work focuses on ensuring safety under the uncertainty in learned human dynamic models.…”

“…In order to ensure safety, a pessimistic robot might assume that all space could be occupied by its human partner in the future, resulting in the freezing robot problem [41]. Preventing harm without unnecessarily impacting task efficiency is a critical challenge [49,28]. In this work, with a focus on interactive manipulation tasks, we address this challenge via a motion planner that guarantees human safety, equipped with a two-pronged definition of safety that infuses flexibility into the planner.…”

HJB-RL: Initializing Reinforcement Learning with Optimal Control Policies Applied to Autonomous Drone Racing

“…For example, in [5] the authors solve the HJB equation for a simplified car model to pass through stochastic gates (similar to a drone race, but in 2D). Similarly, [19] incorporates the solution of an HJ equation for a reduced-order model of relative motion between two cars, to enforce safety within a more traditional MPC planning loop for autonomous driving. The paper [8] presents a method to use an HJB controller derived from a low-order model approximation to safely control the original higherorder system.…”

Safe Controller Synthesis With Tunable Input-to-State Safe Control Barrier Functions

“…The extension of the latter to control barrier functions (CBF) provides a tool for control design by imposing an easy-to-compute condition over a desired safe set. A recent survey on CBFs can be found in [4], and alternative methods for safety-critical control in [5], [6].…”