ARGOS: An Automaton Referencing Guided Overtake System for Head-to-Head Autonomous Racing

TL;DR

The paper tackles the challenge of head-to-head autonomous racing by introducing ARGOS, a modular automaton network that decomposes overtaking and position defense into sequential submaneuvers supervised by three interconnected automatons (ARGOS, AutoPass, KAVAL). It couples perception, a quintic spline-based local planner, MPC tracking, and the AEMS boost mechanism within a formal verification framework to ensure safety and rule compliance. Formal design verification and model checking, aided by counterexamples, validate the framework's behavior and guide refinements, while LGSVL-based experiments demonstrate multiple overtakes and defenses with performance comparable to human-driven racing. The work advances multi-agent autonomous racing by providing a verifiable, flexible architecture that can adapt to different race rules and opponent capabilities, with practical implications for safe, high-speed autonomous competition.

Abstract

Autonomous overtaking at high speeds is a challenging multi-agent robotics research problem. The high-speed and close proximity situations that arise in multi-agent autonomous racing require designing algorithms that trade off aggressive overtaking maneuvers and minimize the risk of collision with the opponent. In this paper, we study a special case of multi-agent autonomous race, called the head-to-head autonomous race, that requires two racecars with similar performance envelopes. We present a mathematical formulation of an overtake and position defense in this head-to-head autonomous racing scenario, and we introduce the Automaton Referencing Guided Overtake System (ARGOS) framework that supervises the execution of an overtake or position defense maneuver depending on the current role of the racecar. The ARGOS framework works by decomposing complex overtake and position-defense maneuvers into sequential and temporal submaneuvers that are individually managed and supervised by a network of automatons. We verify the properties of the ARGOS framework using model-checking and demonstrate results from multiple simulations, which show that the framework meets the desired specifications. The ARGOS framework performs similar to what can be observed from real-world human-driven motor sport racing.

Figures (17)

• Figure 1: A block diagram view of the MPC based path tracker. Using the kinematic bicycle model of the Dallara AV-21 racecar, the path tracker can accurately determine the optimal steering and throttle to track a reference trajectory.
• Figure 2: A timed odometry trace for: [Top] a successful position defense, and [Bottom] a successful overtake attempt. In both cases, the red and green dots represent the opponent and ego racecar respectively. Each set of ego and opponent poses represent the sub-maneuvers involved in an overtake attempt or position defense.
• Figure 3: The piecewise trapezoidal overtake geometry model showing the three submaneuvers involved in an overtake [left-right] (A) initiate overtake, (B) pass opponent's car-length, and (C) safely merge in-front of the opponent.
• Figure 4: An overview of the quintic spline generation process used in this paper to produce a local overtake (red) and position defense splines (green) using the corresponding set of guide control points.
• Figure 5: Geometric view of superprojection intercepting position-defense model. The ego racecar must find a pose in the opponent's superprojection and occupy that pose before the opponent to successfully block the opponent.