画像特徴量による移動式マニピュレータの位置決め制御

抄録

This manuscript reports on a vision-based autonomous grain unloading system for a head-feeding combine robot. In order to perform harvesting fully autonomously, the combine robot is also required to autonomously unload the harvested grain into a wagon through the unloading auger, which is a screw auger and modelled as a 2-DOF manipulator. We presented a vision-based method for the autonomous unloading. A machine vision system was developed that consists of a camera on the combine robot and a planar target on the wagon. Based on SURF, the robot locates the wagon and positions its auger spout at an appropriate point over the wagon. We conducted field experiments to evaluate the presented method. Experimental results showed that the combine robot was able to position its auger spout at the predetermined target point with tolerable accuracy.