ForzaETH Race Stack -- Scaled Autonomous Head-to-Head Racing on Fully Commercial off-the-Shelf Hardware (2403.11784v2)

Abstract: Autonomous racing in robotics combines high-speed dynamics with the necessity for reliability and real-time decision-making. While such racing pushes software and hardware to their limits, many existing full-system solutions necessitate complex, custom hardware and software, and usually focus on Time-Trials rather than full unrestricted Head-to-Head racing, due to financial and safety constraints. This limits their reproducibility, making advancements and replication feasible mostly for well-resourced laboratories with comprehensive expertise in mechanical, electrical, and robotics fields. Researchers interested in the autonomy domain but with only partial experience in one of these fields, need to spend significant time with familiarization and integration. The ForzaETH Race Stack addresses this gap by providing an autonomous racing software platform designed for F1TENTH, a 1:10 scaled Head-to-Head autonomous racing competition, which simplifies replication by using commercial off-the-shelf hardware. This approach enhances the competitive aspect of autonomous racing and provides an accessible platform for research and development in the field. The ForzaETH Race Stack is designed with modularity and operational ease of use in mind, allowing customization and adaptability to various environmental conditions, such as track friction and layout. Capable of handling both Time-Trials and Head-to-Head racing, the stack has demonstrated its effectiveness, robustness, and adaptability in the field by winning the official F1TENTH international competition multiple times.

• The paper introduces a robust autonomous racing stack that leverages COTS hardware to enable head-to-head competition with high-speed precision.
• It details an integrated system combining perception, planning, and control—using EKF-based state estimation, global/local planning, and adaptive opponent tracking.
• Evaluations in Time-Trials and Head-to-Head scenarios demonstrate competitive lap times, reduced lateral deviation, and scalable real-time performance.

Overview of the ForzaETH Race Stack for Autonomous Racing

The paper "ForzaETH Race Stack - Scaled Autonomous Head-to-Head Racing on Fully Commercial off-the-Shelf Hardware" presents a comprehensive solution for autonomous racing using the F1TENTH platform, which is a 1:10 scale racing competition focused on Head-to-Head races. This research delineates the ForzaETH Race Stack, a robotic system designed for high-speed, autonomous racing. It amalgamates several key robotics and AI modules encompassing perception, planning, and control, all executed on commercially available off-the-shelf hardware.

Architecture and Design

The ForzaETH Race Stack is organized following the See-Think-Act paradigm, integrating perception, state estimation, planning, and control components. This structure aligns with established methodologies in autonomous robotics, yet the focus on high-speed racing introduces unique challenges related to latency, the need for robust perception in dynamic environments, and precise state estimation. The hardware setup involves an Intel NUC as the onboard computer, Hokuyo UST-10LX LiDAR, and a VESC motor controller, all powering a modified Traxxas RC vehicle.

Key Components

1. State Estimation: Combines odometry, IMU, and LiDAR data to provide accurate vehicle positioning. It uses the Extended Kalman Filter (EKF) for velocity estimation and offers two localization methods: Cartographer-SLAM and SynPF. The latter offers robustness against wheel slip, a frequent issue in racing scenarios.
2. Opponent Estimation: Employs adaptive clustering on LiDAR data for detecting opponents and uses an EKF for tracking, even when the opponent is out of sight. This module is crucial for maintaining a competitive edge during races by ensuring reliable detection and tracking crucial for overtaking maneuvers.
3. Planning: Includes a global planner, which computes a racing line using a minimum curvature optimization technique, and a local planner, which adapts the trajectory for overtaking opponents. The system uses a sector-based tuning method to adjust speed profiles, optimized via Bayesian optimization to balance lap time and safety.
4. Control: The Race Stack utilizes the Model- and Acceleration-based Pursuit (MAP) controller for lateral control and a PD controller for longitudinal control during overtaking. This combination allows precise trajectory tracking and vehicle stability at high speeds.

Performance and Evaluation

The Race Stack's performance is evaluated in both Time-Trials and Head-to-Head scenarios. During Time-Trials, the stack aims to maximize consecutive crash-free laps while achieving fast lap times, balancing robustness and speed. In Head-to-Head races, the emphasis shifts to strategic positioning against opponents, utilizing the planning and opponent estimation modules to perform safe and efficient overtakes.

1. Time-Trials: The stack shows robust consistency in lap times, with careful tuning of racing line velocity using sector scalers optimizing performance. The model demonstrates a superior lap time and reduced lateral deviation compared to other known control approaches like Pure Pursuit (PP) and Follow-The-Gap (FTG).
2. Head-to-Head Racing: The ForzaETH Race Stack showcases its ability by maintaining a competitive racing pace against varying opponent configurations, employing the overtaking planner to successfully navigate slower opponents based on available speed advantages.

Computational Efficiency

The computational demand on the onboard Intel i5-10210U CPU is carefully managed, with the state estimation modules being the most resource-intensive. The system optimization ensures that the computational overhead remains within acceptable bounds to maintain real-time performance, crucial for racing contexts.

Implications and Future Work

The implications of this work extend beyond autonomous racing to general real-time, high-speed robotic applications. The modular design on COTS hardware lowers the barrier for research in high-speed robotics, fostering innovation and further developments in autonomous vehicle research. Future research directions include expanding the system to handle multi-opponent races and exploring its scalability to full-sized autonomous vehicles, potentially broadening its application scope within autonomous driving research.

This paper successfully details how a fully functional, competitive, and accessible autonomous racing stack can be achieved in a resource-constrained setting while paving the way for future advancements and applications in the field of high-speed autonomous systems.