Navigating Narrow Spaces: A Comprehensive Framework for Agricultural Robots
Geesara Kulathunga, Abdurrahman Yilmaz, Zhuoling Huang, Ibrahim Hroob, Leonardo Guevara, Jaspreet Singh, Grzegorz Cielniak, Marc Hanheide
AI summary
Problem
Autonomous robots struggle to navigate narrow, cluttered agricultural spaces due to weak GNSS signals, sensor noise, and kinematic constraints, leaving a gap in reliable, low-localization-accuracy navigation methods.
Approach
The authors propose a modular framework that processes real-time LiDAR point clouds to detect boundary poles and estimate centerlines, then refines initial paths via non-linear optimization while tracking them with a speed-adaptive pure pursuit controller.
Key results
- Modular perception-driven navigation framework for constrained environments
- Robust PCA-based pole detection and boundary estimation pipeline handling occlusions
- Non-linear optimization trajectory refinement ensuring kinematic and spatial constraint adherence
- 0.08 ± 0.01 m average lateral deviation in real polytunnel tests, outperforming RTEB and MPPI
Why it matters
Provides a reliable, generalizable navigation solution for agricultural robots operating in GNSS-denied, narrow-row environments, directly supporting precision farming automation.
Abstract
Navigating within narrow spaces is a fundamental challenge in robotics, requiring precise localisation, localisation error recovery, dynamic path planning, and adaptive con- trol for effective manoeuvring. This paper presents a modu- lar and perception-driven navigation framework designed for constrained environments, focusing primarily on agricultural applications. The proposed method integrates a multi-step point cloud processing pipeline for robust local perception, includ- ing pole detection, boundary line estimation, and trajectory refinement to ensure safe and precise traversal by refining initial trajectories based on detected environmental constraints and dynamically adapting to kinematic limitations. Experimen- tal validation in a real strawberry polytunnel demonstrates superior trajectory accuracy and control stability compared to state-of-the-art navigators, achieving an average lateral deviation of 0.08 ± 0.01 m. The adaptive trajectory tracking and regulated pure pursuit control of the framework contribute to consistent navigation, even under increased velocity constraints, outperforming the resilient timed elastic band (RTEB) and model predictive path integral (MPPI) methods. This modular and generalisable framework offers significant potential for advancing autonomous navigation in narrow-space applications.