A consistent heuristic for efficient path planning on mobility maps

Abstract

Efficient path planning has an utmost importance in the domain of autonomous navigation. Even though shortest path planning has been well discussed in the past, the same techniques might not always be used to find the fastest path in outdoor environments due to the inability of mobile agents to travel at their peak speed everywhere in irregular terrains. Mobility maps are an effective way for dealing with such irregularities. In this paper, we first introduce a grid-based mobility maps for representing speed limitations in outdoor terrains. Then, we propose a heuristic for finding the fastest path on such maps. The proposed heuristic is proven to be both admissible and consistent. Therefore, it can be used with A*-like heuristic search algorithms for obtaining fastest paths efficiently. Simulation results provided in this paper verify the optimality of paths that are found with the help of the proposed heuristic.

Path-Planning-Enabled Semiflocking Control for Multitarget Monitoring in Mobile Sensor Networks

“…Anti-flocking in [23] Semi-flocking in [17] Semi-flocking in [18] Proposed semi-flocking (a) Average instantaneous area coverage (%)…”

“…In this paper, a time heuristic is proposed for finding the fastest path on mobility maps efficiently. The proposed time heuristic is proven to be both admissible and consistent [23]. When an A* search algorithm is guided by the proposed time heuristic, it guarantees an optimal solution can be found if there is such a solution for the problem.…”

“…There are many factors that dominate the maximum allowed or achievable speed between any two points on a given AoI. Examples include [23] • the availability of traction to overcome resistances resulting from terrain roughness, terrain slope, blocking obstacles, and vegetation density, etc; and • the maneuverability of the node across obstacles. Since this work focuses on incorporating path planning for navigation in the semi-flocking-controlled nodes, the construction of the mobility map is regarded as out of scope.…”

“…Four different algorithms are put under test for path planning on mobility maps. They are (1) Dijkstra's algorithm [23], (2) A* algorithm [23], (3) A* algorithm with path lookup (A*PL), and (4) A* with intersection lookup (A*IL). Both the Dijkstra and A* algorithms are in their generic versions for generating time-efficient paths between the given start and end points on a mobility map.…”

“…Referring to Table I, Dijkstra's algorithm has expanded 16157 nodes using 235.66 seconds in its search for the fastest path whereas A* with the proposed heuristic has found the exact same path by just expanding 14751 nodes using 221.03 seconds. With the help of heuristic functions to estimate the time-cost to the end point [23], the A*-based methods can find paths faster than the Dijkstra's method. While the generic A* algorithm improves only slightly over the Dijkstra's algorithm in terms of execution time, both A*PL and A*IL algorithms show significant improvements with the use of static routing tables.…”