There appears to be a division between control-centric and planning-centric approaches to problems like the APC. The two top solutions made extensive use of visual servoing and force control. They also did not involve “grasping”, i.e., explicit reasoning regarding grasping poses, and only a very limited amount of ”motion planning”, i.e., collision free planning in a configuration space representation. Across other teams, visual servoing and force control techniques were sparsely used, and furthermore only two teams identified these techniques to have been absent in their approaches. By contrast, only four teams outside of the top two failed to implement either grasping or motion planning. Although it is tempting to draw general conclusions based on this data, the third-place team (Grizzly) did not use any reactive control schemes, but relied on motion planning and SLAM (Hector SLAM ROS package [44]) to localize and move a mobile Baxter robot in front of the appropriate bin. It is therefore unclear what the “best” approach is, albeit the strategy of using reactive control to compensate for inaccurate sensing and actuation of an underlying deliberative architecture appears to be powerful.