A Modular Architecture Design for Autonomous Driving Racing in Controlled Environments

TL;DR

The paper presents a modular architecture for autonomous racing in controlled environments, integrating perception, localization/mapping, trajectory planning, and control within a ROS2-based pipeline. It implements a dual-computer hardware design with RTK-enabled positioning, a ZED 2i stereo camera, and a YOLOv11s-based cone detector, coupled with EKF-based mapping and spline-based path planning to enable real-time navigation. Experimental results show improved localization with RTK, quantifiable depth estimation characteristics, and successful simulation-based end-to-end validation of the control pipeline, supporting reliable track following and safety gating. The approach demonstrates a practical, scalable framework for autonomous racing in constrained tracks, balancing computational load with real-time requirements.

Abstract

This paper presents an Autonomous System (AS) architecture for vehicles in a closed circuit. The AS performs precision tasks including computer vision for environment perception, positioning and mapping for accurate localization, path planning for optimal trajectory generation, and control for precise vehicle actuation. Each subsystem operates independently while connecting data through a cohesive pipeline architecture. The system implements a modular design that combines state-of-the-art technologies for real-time autonomous navigation in controlled environments.

Figures (10)

• Figure 1: Hardware Architecture
• Figure 2: Software Architecture
• Figure 3: Kinematic Bicycle Model (Tricycle representation)
• Figure 4: Cone ordering algorithm.
• Figure 5: Object detection F1 curve and PR curve.