Automated Test Case Generation for BDI Agents

摘要

We propose a coverage oriented test case generation methodology for BDI multi-agent systems. The coverage criteria involve plans and nodes within plans of multi-agent systems. We organise the criteria into a subsumption hierarchy to show the coverage relationships between the criteria. Then we apply the criteria on multi-agent systems to analyse some empirical data. The data analysed is the effect on the number of test cases generated automatically for each criterion. We use a tool, BDITESTER, to obtain the empirical data and to show that our proposal is pragmatic. Finally, we suggest some guidelines to select a criterion to automatically generate test cases for BDI agents.