Combination and mutation strategies to support test data generation in the context of autonomous vehicles

摘要

The software used to control autonomous vehicles is a type of embedded system that needs to undergo strenuous testing before deployment. Field testing is the final stage of testing ensuring that autonomous vehicles show the intended behaviour. It usually does not take into consideration the code structure. In this context, a previously proposed testing model and a software tool to support structural testing in the context of autonomous vehicle field testing have been improved to support the generation of new input data from logs collected during field testing using strategies of combination and mutation. We present in this paper three combination strategies and five mutation strategies with the objective of being used in a search-based algorithm for structural data testing generation. A study to assess their coverage according to the criteria all-nodes and all-edges is also shown.