PhD studentship: Assured and Scalable Self-Adaptation for the Engineering of Trustworthy Autonomous Robotic Teams

Robotic teams and other distributed autonomous systems (DAS) are increasingly used in several application domains including logistics, manufacturing, and infrastructure inspection. Assuring the trustworthiness of DAS carrying out a safety-critical task collaboratively is very challenging due to uncertainties and risks associated with the operating environment, team member failures, etc. To overcome these challenges, the distributed-control software of DAS should exhibit high levels of scalability and optimality underpinned by assurance evidence that the DAS operates safely in continually changing and unexpected scenarios. Existing DAS approaches satisfy only a subset of these key characteristics of trustworthy DAS: Data-driven approaches using machine learning albeit scalable and capable to produce close-to-optimal solutions lack the guarantees needed in safety-critical tasks (Sykes et al 2011, Nallur et al 2013, Grassi et al 2013), while model-based approaches using formal verification (Hunter et al 2013) suffer from scalability issues when using large-scale and complex models.

The PhD project will contribute significantly to addressing this challenge by leveraging the capabilities from both data-driven (using machine learning) [5,6] and model-based (using formal methods) [7] paradigms to devise assured and scalable self-adaptation techniques that support the development of trustworthy distributed-control software for DAS. These techniques will be integrated with state-of-the-art DAS middleware [5] and their feasibility will be validated through a demonstrator both in simulation [8] and using mobile robots available in our lab.

Project enquiries 

 Dr Simos Gerasimou 

 [Email Address Removed] 

 Dr Javier Camara Moreno 

 [Email Address Removed]