The Autonomous Software Stack of the FRED-003C: The Development That Led to Full-Scale Autonomous Racing

TL;DR

The paper presents the autonomous software stack developed by BME Formula Racing Team (FRED-003C) for Formula Student Driverless competitions and its progression toward full-scale autonomous racing in A2RL. It describes a modular pipeline with state estimation, perception, planning, and control implemented on a two-layer hardware architecture, detailing methods, experiments, and race results that motivated sharing. The main contributions include Fast SLAM-based state estimation, LiDAR–camera perception with robust fusion, a centerline-based planning approach with smoothing and velocity profiling, and a hybrid lateral controller that combines Stanley and Pure Pursuit. The work demonstrates a practical, high‑performance autonomous racing blueprint suitable as a starting point for student teams and highlights its educational and industry impact.

Abstract

Scientific development often takes place in the context of research projects carried out by dedicated students during their time at university. In the field of self-driving software research, the Formula Student Driverless competitions are an excellent platform to promote research and attract young engineers. This article presents the software stack developed by BME Formula Racing Team, that formed the foundation of the development that ultimately led us to full-scale autonomous racing. The experience we gained here contributes greatly to our successful participation in the Abu Dhabi Autonomous Racing League. We therefore think it is important to share the system we used, providing a valuable starting point for other ambitious students. We provide a detailed description of the software pipeline we used, including a brief description of the hardware-software architecture. Furthermore, we introduce the methods that we developed for the modules that implement perception; localisation and mapping, planning, and control tasks.

Figures (14)

• Figure 1: The perception sensor placement of the FRED-003C. The LiDAR sensor is placed on the nose cone and the camera equipped with a wide angle lens is mounted to the top of the main hoop.
• Figure 2: Block diagram depicting the hardware-software architecture of the autonomous system. The software pipeline can be divided into four parts; state estimation, perception, planning, and control. The execution is distributed between a high-performance compute layer and a real-time layer.
• Figure 3: LiDAR point cloud segmented into ground points marked by blue, non ground points marked by red, and classified cones marked by green.
• Figure 4: LiDAR cluster centers marked by green crosses are projected onto the image containing bounding boxes of the detected cones.
• Figure 5: Delaunay triangulation, where yellow and blue triangles represent the cones that bound the track, and the dashed lines show the resulting partitioning.