Cognitive factories with multiple teams of heterogeneous robots: Hybrid reasoning for optimal feasible global plans

Abstract

We consider cognitive factories with multiple teams of heterogenous robots, and address two key challenges of these domains, hybrid reasoning for each team and finding an optimal global plan (with minimum makespan) for multiple teams. For hybrid reasoning, we propose (i) modeling each team's workspace taking into account capabilities of heterogeneous robots, (ii) embedding continuous external computations into discrete symbolic representation and reasoning by combining different methods of integration, (iii) not only optimizing the makespans of local plans but also minimizing the total cost of robotic actions, where costs of actions can be defined in various ways. To find a global plan with minimum makespan, we propose a semi-distributed approach: we formulate the problem of finding an optimal coordination of teams that can help each other, prove its intractability, and describe how to solve this problem using existing automated reasoners. As a case study, we show applications of our hybrid reasoning and coordination approaches on a cognitive toy factory with dynamic simulations and physical implementation utilizing KuKa youBots and Lego NXT robots (supplementary video provided). We also present experimental results to discuss the scalability of these methods.