A penalty search algorithm for the obstacle neutralization problem

Ali Fuat Alkaya, Vural Aksakalli, Carey E. Priebe
2015 Computers & Operations Research  
We consider a path planning problem wherein an agent needs to swiftly navigate from a source to a destination through an arrangement of obstacles in the plane. We suppose the agent has a limited neutralization capability in the sense that it can safely pass through an obstacle upon neutralization at a cost added to the traversal length. The agent's goal is to find the sequence of obstacles to be neutralized en route that minimizes the overall traversal length subject to the neutralization
more » ... We call this problem the obstacle neutralization problem (ONP), which is essentially a variant of the intractable weight-constrained shortest path problem in the literature. In this study, we propose a simple, yet efficient and effective suboptimal algorithm for ONP based on the idea of penalty search and we present special cases where our algorithm is provably optimal. Computational experiments involving both real and synthetic naval minefield data are also presented. tation of at most K of the disks for neutralization. Clearly, ONP is a variant of the WCSPP where the weight constraint is the number of neutralizations available. On the other hand, to our knowledge, ONP as defined above has not been studied before in the open literature except for a brief mention in [5] . Figure 1 with s = (10, 20) , t = (10, 1), and C = 0.8 with r = 3. In the figure, the optimal paths for K = 0, 1, 2 and 3 are given in subfigures 1(a), 1(b), 1(c), and 1(d), respectively. Note that for K = 1, only d 6 is neutralized; for K = 2, both d 5 and d 6 are neutralized; and for K = 3, d 5 , d 6 , and d 7 are neutralized. Note also that it is not always the case that all K allowable neutralizations are used. An instance of ONP is shown in Our goal in this study is to present a simple, fast, efficient, and scalable algorithm for ONP which we call Penalty Search Algorithm (PSA). This algorithm is especially suitable for online applications. State-of-the-art algorithms
doi:10.1016/j.cor.2014.08.013 fatcat:f3oyn3mlhvgwfi2hndr2y3zspi