M-Predictive Spliner: Enabling Spatiotemporal Multi-Opponent Overtaking for Autonomous Racing

TL;DR

This paper introduces M-Predictive Spliner, a novel method enabling autonomous racing cars to perform multi-opponent overtaking by predicting opponents' future trajectories, significantly improving safety and success rates in physical experiments.

Contribution

It presents a new spatiotemporal decision-making approach for multi-opponent racing that incorporates opponent intent prediction and trajectory forecasting.

Findings

Achieved up to 91.65% overtaking success rate

Improved safety by 10.13 percentage points over state-of-the-art

Validated on a physical 1:10 scale autonomous racing car

Abstract

Unrestricted multi-agent racing presents a significant research challenge, requiring decision-making at the limits of a robot's operational capabilities. While previous approaches have either ignored spatiotemporal information in the decision-making process or been restricted to single-opponent scenarios, this work enables arbitrary multi-opponent head-to-head racing while considering the opponents' future intent. The proposed method employs a KF-based multi-opponent tracker to effectively perform opponent ReID by associating them across observations. Simultaneously, spatial and velocity GPR is performed on all observed opponent trajectories, providing predictive information to compute the overtaking maneuvers. This approach has been experimentally validated on a physical 1:10 scale autonomous racing car, achieving an overtaking success rate of up to 91.65% and demonstrating an average 10.13%-point improvement in safety at the same speed as the previous SotA. These results highlight its potential for high-performance autonomous racing.