Comparison of A* and RRT in real–time 3D path planning of UAVs

Abstract

Unmanned Aerial Vehicles (UAVs) are being integrated into a wide range of military, industrial and commercial applications. Such applications require faultless autonomous systems to coordinate, guide, navigate and control different UAVs of different sizes, designed for different purposes with different capabilities. In this regard, different path planning algorithms were developed to ensure that UAVs are supplied with collision-free path paramount to which are the A* and the RRT algorithms, a graph-based and a samplingbased algorithm respectively. Such algorithms shall ideally operate in real-time to furnish the UAV navigation system with real-time, valid, obstacle-free paths in view of changes in the environment or other external or user-defined restrictions. Owing this need, in this paper a real-time platform to assess the performance of the A* and RRT algorithm with an associated smoothing algorithm was developed and tested using 3, 3D obstacle environment with different complexities. The salient user-defined, system-defined and internal constants were independently considered and their effect on performance assessed. Results showed that the A* outperformed the RRT algorithm in both path length and computational time for all scenarios considered with difference increasing with scenario complexity. But, both algorithms can be utilised if the associated parameters are attentively chosen based on the scenario the UAV will operate as both algorithm reached a 100% success rate for all scenario at specific parameter assignments.

3D real-time path planning of UAVs in dynamic environments in the presence of uncertainty

“…[76][77][78] Just as for the A* algorithm, a more detailed explanation of the RRT algorithm and its variants is provided our previous works. [11][12][13]…”

“…A more detailed explanation of this algorithm is provided in our previous works. 12,13 This post-path construction smoothing algorithm was developed to mitigate with the non-optimality characteristics of the RRT algorithm. Since in our previous works, 11,12 the A* algorithm produced the shortest possible path with the considered resolution, improvement with the smoothing algorithm was only marginal.…”

“…[79][80][81] Owing that the scope of this work is to design a path planning algorithm for such scenarios, an algorithm to emulate real-time behaviour was developed in a previous paper. 13 The developed real-time path planning algorithm selects an obstacle-free intermediate goal node in the direction of the final goal position a look-ahead distance from the current UAV position. This look-ahead distance is governed by the sensory system's FOV.…”

“…A no solution situation exists when either a no intermediate obstacle-free path was not constructed in the allocated time and/or the total time to reach the goal from the initial position has exceeded. The various UAV parameter definitions defined in our previous work, 13 are briefly explained in Section IV C. For a more detailed explanation of the described real-time algorithm and performance results for both path planning algorithms, refer to our previous work. 13…”

“…Section III provides a brief resume of the A* and RRT algorithms, the smoothing algorithm which is only applied to the RRT algorithm, the real-time path planning platform and the dynamic obstacle definition framework all extensively described in our previous works. [10][11][12][13] Section IV will define the environmental scenarios, the UAV model and path planner parameters definitions and constraints and uncertainty modelling and quantification rationale in view of 3D real-time path planning algorithm. Section V will present and analyse the results in view of real-time 3D UAV path planning requirements.…”