A large proportion of the work carried out in Semantic Robots’ research profile is connected to some form of automated planning, that is, deciding what robots should do, and when, where, and how they should do it. The case studies considered by Semantic Robots require the development of techniques for several forms of automated planning, namely, multi-vehicle coordination, task allocation, motion planning for non/holonomic robots, and robot control. Crucially, many case studies require all these problems to be resolved jointly. This is often called hybrid problem solving.

General techniques for hybrid problems is an open issue in research, and as of yet there is no general theory of hybrid problem solving. Researchers at Semantic Robots are exploring the theory and practice of hybrid problem solving in this general setting. We believe that understanding this problem in its most general terms will lead to innovations that will enable the development of specific solutions for particular instances of the general problem.

Our industrial partners contribute significant variants of the general problem (see Case Studies of Atlas Copco, Kollmorgen, and Volvo CE) that we are using to test our general techniques. Semantic Robots’ researchers are developing a theoretical framework for hybrid problem solving based on the so-called Meta-CSP approach. The approach is grounded in the Meta-Constraint Satisfaction Problem and a software framework that exemplifies how the technique can be used is under development. The entire framework is publicly available at metacsp.org.