Application of metamorphic testing on UAV path planning software

Abstract

Both the performance and reliability evaluation processes of the unmanned aerial vehicle path planning software rely on the determination of the correctness of the execution results of unmanned aerial vehicle path planning software. However, this task is hindered due to the testing oracle problem. In this paper, a framework is designed to overcome the oracle problem and verify the correctness of path planning software based on the grid searching algorithms. In this framework, a metamorphic testing-based method is proposed, and three operations-based metamorphic relations are proposed and proved towards the target software. While analysis of the software is conducted, the version with manually injected faults as well as the officially released version are both dealt with. It is shown that in the experimental results the injected faults can be effectively revealed by using the methods proposed in this paper. Besides, through the evaluation of different types of metamorphic relations, we find that the composed metamorphic relations have stronger fault detection capability compared to the individual ones.