Approximation Algorithms for Pick-and-Place Robots

Abstract

In this paper we study the problem of finding placement tours for pick-and-place robots, also known as the printed circuit board assembly problem with m positions on a board, n bins containing m components and n locations for the bins. In the standard model where the working time of the robot is proportional to the distances travelled, the general problem appears as a combination of the travelling salesman problem and the matching problem, and for m=n we have an Euclidean, bipartite travelling salesman problem. We give a polynomial-time algorithm which achieves an approximation guarantee of 3+ε. An important special instance of the problem is the case of a fixed assignment of bins to bin-locations. This appears as a special case of a bipartite TSP satisfying the quadrangle inequality and given some fixed matching arcs. We obtain a 1.8 factor approximation with the stacker crane algorithm of Frederikson, Hecht and Kim. For the general bipartite case we also show a 2.0 factor approximation algorithm which is based on a new insertion technique for bipartite TSPs with quadrangle inequality. Implementations and experiments on “real-world” as well as random point configurations conclude this paper.