Integrating LiDAR-Based SLAM and Human-Aware Predictive Navigation for Optimal AGV Path Planning

Our project develops a LiDAR-based autonomous navigation system for Orlar Vietnam's indoor farm AGV, replacing its current magnetic-line guidance with a self-localizing, obstacle-aware platform. The existing AGV has no onboard sensing at all: it can only follow a single fixed magnetic track, cannot detect obstacles, and depends on continuous manual operator input, leaving no way to sense or avoid people, crop tables, or other objects in the aisle. 

To resolve this, the AGV is upgraded with an onboard LiDAR and a pre-built map of the farm, replacing the physical track with a fixed, pre-defined route it can localize itself against. As it drives, the LiDAR continuously feeds a local cost map, letting the path-following controller steer smoothly around unexpected obstacles including people in the aisle while staying within its safe corridor and never entering restricted areas such as underneath crop tables. An onboard camera captures crop images at intervals as the AGV passes each row. 



By replacing an unguided, track-bound AGV with one that senses and reacts to its surroundings in real time, this project removes the reliance on rigid physical infrastructure and constant manual supervision, improving both the safety and reliability of autonomous inspection laying the groundwork for Orlar Vietnam's broader goal of reliable, sensor-driven automation across its indoor farming operations.