Answers are grounded in this paper and cite their sources.
Small-sized autonomous agricultural machinery is well suited to narrow and fragmented upland fields, where reliable ridge navigation is required under constraints in payload capacity, computational resources, energy supply, and cost. Ridge tops, ridge edges, furrows, and crop rows can provide local navigation references, although their visibility varies with crop growth and field conditions. This review integrated ridge detection and ridge-following technologies across the navigation chain, from field perception and reference generation to path tracking, steering control, headland transition, and ridge re-entry. Literature was primarily identified through targeted Google Scholar searches, supplemented by structured searches in Scopus and Web of Science to broaden coverage and cross-check relevant publications. Vision-based approaches, including conventional image processing, stereo vision, and deep learning, were reviewed together with light detection and ranging (LiDAR), GNSS/RTK-GNSS, ultrasonic sensing, and multisensor fusion. Perception and control were evaluated jointly because navigation reliability depends on detection error, processing latency, vehicle slip, actuator response, operating speed, and row-transition performance. The literature shows that no single sensor or controller is universally suitable across crop types, ridge geometries, growth stages, and field conditions. Key gaps include limited cross-season datasets, inconsistent performance metrics, insufficient embedded-platform and energy evaluation, weak perception-control co-design, and limited multi-ridge validation. Future work should prioritize standardized benchmarks, lightweight multimodal perception, adaptive sensor fusion, kinematics-matched control, and long-duration field validation.
No takes yet. Share an insight, caveat, or question.
Tu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: