Rapid Development and Performance Evaluation of a Potato Planting Robot

Agriculture is characterised by noisy semi-structured conditions, and production uncertainty which present obstacles to automation. This paper introduces PotatoBot, a specialised robot designed for planting potatoes in a small niche market with limited resources. By adopting an agile-like approach and integrating off-the-shelf modular components, the agri-robotics prototype was developed rapidly and cost-effectively. The system combines Mask R-CNN and PCA for potato pose localisation and utilises a custom end-effector for efficient grasping within standard box crates. Extensive metrics, farmer engagement, and laboratory simulations were used to evaluate performance. Our approach enables the rapid development of an affordable robotic system, particularly suited for niche farming markets with tight profit margins.