A general constraint-based programming framework for multi-robot applications

Abstract

Classic task programming methods based on the specification of desired Cartesian frames can easily generate overconstrained task specifications, reducing the motion capabilities of the involved robot(s) and increasing the total programming effort. This paper presents a general constraint-based programming framework for the specification of a task as minimum set of constraints and the automatic generation of motion optimization problems. The framework can handle constraints involving both robot joint and Cartesian coordinates, as well as including explicit time dependency. The proposed formalism naturally scales to robotic applications with multiple robots, on which multiple frames might be of interest. Additionally, the paper proposes a theoretical comparison with already existing constraint-based programming methods. Finally, the validity and the effectiveness of the proposed approach is numerically supported by illustrative examples, as well as by case studies mocking real industrial setups.