End2Race: Efficient End-to-End Imitation Learning for Real-Time F1Tenth Racing

2025-09-20

AI Summary

To address real-time and robustness challenges in end-to-end imitation learning for the F1Tenth platform under high-speed, dynamic, and head-to-head interaction scenarios, this work proposes a lightweight and efficient architecture. It employs a Gated Recurrent Unit (GRU) to model temporal driving decisions and introduces a novel “spatial pressure token” representation: raw LiDAR scans are normalized via Sigmoid to encode both local perceptual fidelity and global policy continuity. This design enhances model capacity and training stability while achieving sub-0.5 ms inference latency on consumer-grade GPUs. Evaluated across 2,400 high-density overtaking simulation scenarios, the method attains a 94.2% safety rate and a 59.2% successful overtaking rate—surpassing state-of-the-art end-to-end approaches in key metrics. Results demonstrate its comprehensive advantages in real-time performance, operational safety, and competitive maneuver generation.

Technology Category

• Machine Learning: Imitation Learning & Inverse Reinforcement Learning
• Planning, Routing, and Scheduling: Planning with Language Models
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems

Abstract

F1Tenth is a widely adopted reduced-scale platform for developing and testing autonomous racing algorithms, hosting annual competitions worldwide. With high operating speeds, dynamic environments, and head-to-head interactions, autonomous racing requires algorithms that diverge from those in classical autonomous driving. Training such algorithms is particularly challenging: the need for rapid decision-making at high speeds severely limits model capacity. To address this, we propose End2Race, a novel end-to-end imitation learning algorithm designed for head-to-head autonomous racing. End2Race leverages a Gated Recurrent Unit (GRU) architecture to capture continuous temporal dependencies, enabling both short-term responsiveness and long-term strategic planning. We also adopt a sigmoid-based normalization function that transforms raw LiDAR scans into spatial pressure tokens, facilitating effective model training and convergence. The algorithm is extremely efficient, achieving an inference time of less than 0.5 milliseconds on a consumer-class GPU. Experiments in the F1Tenth simulator demonstrate that End2Race achieves a 94.2% safety rate across 2,400 overtaking scenarios, each with an 8-second time limit, and successfully completes overtakes in 59.2% of cases. This surpasses previous methods and establishes ours as a leading solution for the F1Tenth racing testbed. Code is available at https://github.com/michigan-traffic-lab/End2Race.

Problem

• Developing efficient algorithms for autonomous racing with high-speed decision-making constraints
• Addressing challenges of head-to-head racing in dynamic environments requiring strategic planning
• Overcoming limitations in model capacity for real-time autonomous racing applications

Innovation

• Gated Recurrent Unit architecture for temporal dependencies
• Sigmoid-based normalization for LiDAR scan processing
• Sub-millisecond inference time on consumer GPU