Project background
Outdoor autonomous vehicles — mowers, sprayers, inspection robots — share a need for efficient coverage and transit planning. The client wanted a reusable planning engine they could embed across platforms.
Challenge
Handling arbitrary field shapes, obstacles, slope constraints, and no-go zones while producing paths that are efficient, safe, and human-intuitive. The engine had to replan quickly when the map changed during execution.
Approach & solution
We implemented a layered planner: a global coverage solver produces initial patterns, a local planner handles obstacles and dynamic changes, and a supervisor stitches the two together. All components are deterministic and heavily tested against synthetic and real-world scenarios.
Results & benefits
The engine now powers planning across several robotic platforms with consistent behavior. Replanning latency stays within the window needed for smooth motion, and operators describe the generated paths as predictable and easy to supervise.
