Fast computation of optimal paths using a parallel Dijkstra algorithm with embedded constraints

• Published in: Neurocomputing (Amsterdam) Vol. 8; no. 2; pp. 195 - 212
• Main Authors: Solka, Jeffrey L., Perry, James C., Poellinger, Brian R., Rogers, George W.
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 1995
• Subjects: Artificial neural system; Optimal path; Route planning; Turning constraints; Turning constraints; Route planning; Artificial neural system; Optimal path
• ISSN: 0925-2312; 1872-8286
• DOI: 10.1016/0925-2312(94)00018-N

We have developed a new optimal path algorithm in which the paths are subjected to turning constraints. The restriction which we have incorporated is the next link in the path must not make an angle exceeding 45 ° in magnitude with the preceeding link. This algorithm has a natural implementation as an artificial neural system with either synchronous or asynchronous weight updating, and as an automata executing on a massively parallel array processor. At a given step in the path solution process our path planning artificial neural system keeps track of all constrained optimal paths flowing into the nodes of the network. This new algorithm has applications to any path planning problem where the vehicle traveling the path is subject to a limited turning capability. The ability of the network to solve for constrained paths is illustrated with both a graph theoretic example and a scenario involving an unmanned vehicle that must travel a constrained path through a real terrain area containing artificially generated keep out zones.