Vision based crop row navigation under varying field conditions in arable fields

Rajitha de Silva, Grzegorz Cielniak, Junfeng Gao

Abstract: Highlights•A novel multi-crop dataset for crop row detection with Sugar Beet and Maize crops.•The field variations in a dataset dictates the crop-agnostic nature of predictions.•The triangle scan method leads to stable navigation, unaffected by crop row mask IoU.•The controller guided the robot precisely despite initial position errors up to 20°.•The EOR detector senses the robot’s arrival at a crop row’s end, guiding its exit.