Details

Autor(en) / Beteiligte

• Eder, Kerstin
• Hierons, Rob
• Dongol, Brijesh
• Timmis, Jon
• Cavalcanti, Ana
• Woodcock, Jim

Titel

Gaining Confidence in the Trustworthiness of Robotic and Autonomous Systems

Ist Teil von

• Software Engineering for Robotics, 2021, p.139-164

Ort / Verlag

Switzerland: Springer International Publishing AG

Erscheinungsjahr

2021

Quelle

Alma/SFX Local Collection

Beschreibungen/Notizen

• Because no single technique is adequate to cover a whole system in practice, a variety of complementary techniques will be needed to gain confidence in the trustworthiness of robotic and autonomous systems. In this chapter, we argue that demonstrable trustworthiness can be achieved by design, through transparency and by rigorous verification and validation. We then concentrate on ensuring correctness during system development, with specific focus on how to achieve implementations that are free from runtime errors and how design flaws can be detected prior to implementation. Our discussion is illustrated on two examples, robot navigation code and Simulink controllers. In both cases, the power of fully automatic theorem proving combined with test-based techniques can deliver strong confidence in system correctness. Attention is then turned to simulation-based verification, focusing in particular on the test generation challenge. We motivate the benefits of introducing agency into the test environment and show how exploiting multi-agent systems for test generation can lead to robust tests that achieve a high level of coverage as part of a coverage-driven verification methodology. Finally, some challenges for robotic and autonomous system verification are discussed, including specification, (more) automation, discipline, innovation and creativity in combining techniques and exploiting AI for verification.