In Preparation for Joint Action

This paper recommends extensive, human-like preparation for joint action. Given the substantive differences in the ways that people and robots perceive and reason, it argues that this knowledge be shared explicitly. Moreover, it assigns responsibilities to each agent based on their strengths and weaknesses. The approach detailed here deliberately explores to acquire knowledge about the specific task it confronts and the skills of the agents in that setting. It then exploits that knowledge to support performance.