Curb-Tracker: An Integrated Curb Following System for Autonomous Vehicles
Jiahao Liang, Yuanzhe Wang, Guohao Peng, Zhenyu Wu, Danwei Wang
AI summary
Problem
Existing curb-following systems suffer from unreliable detection due to varied curb dimensions and roadside interference, while their motion controllers prioritize tracking accuracy over task completion efficiency, often requiring labor-intensive manual pre-mapping.
Approach
The system uses a 2.5-D elevation map with online parameter tuning for robust curb detection, paired with Model Predictive Contouring Control (MPCC) to dynamically optimize both lateral tracking accuracy and longitudinal travel speed.
Key results
- Adaptive curb detection algorithm leveraging 2.5-D elevation mapping with dynamic parameter adjustment
- MPCC-based motion generator balancing precise lateral offset tracking with high task completion efficiency
- Successful real-world implementation and validation on a Hunter 2.0 Ackerman-steering mobile robot
- Demonstrated robustness and adaptability across diverse simulated and physical road environments
Why it matters
Provides a fully autonomous, real-time curb-following solution that eliminates manual pre-mapping and accelerates deployment for autonomous road sweeping and environmental service vehicles.
Abstract
Curbfollowingisacriticaltechnologyforautonomous road sweeping vehicles. However, existing solutions face two pri- mary challenges: 1) unreliable curb detection; and 2) inefficient mo- tion generation. Unreliable curb detection stems from the wide vari- ability in curb dimensions and types, as well as interference from roadside features, such as vegetation and infrastructure. Inefficient motion generation occurs when existing methods prioritize tracking accuracy while neglecting task completion efficiency, leading to prolonged operation times. To address these challenges, we pro- pose Curb-Tracker, an integrated curb-following system designed for autonomous vehicles operating in diverse road environments. First, we develop a robust and adaptive curb detection algorithm that leverages a 2.5-D elevation map of the local environment and dynamically adjusts key parameters online to ensure reliable detection across varying scenarios. Second, to achieve accurate and efficient curb-aligned motion generation, we leverage model predictive contouring control as a tailored framework specifically designed for the curb-following task to generate an optimal control sequence for the vehicle to maintain a specified lateral offset from the curb while maximizing travel progress along it. The proposed system has been implemented on a Hunter 2.0, a front-wheel Ackerman-steering mobile robot, and has been validated through extensive experiments in both Gazebo simulation and real-world environments. Experimental results demonstrate the effectiveness, adaptability, and robustness of the proposed system across a wide range of road scenarios.