Reasonable Suspicion of Misbehaviour of Aerial Platforms

Abstract:  
Aerial Platforms/drones are set to penetrate society across transport, manufacturing, and smart living at an exponential rate. However, this is affecting our capacity and understanding of how-to police them in a wider societal framework. Whilst there are current legal criteria based on physical cause and effect interpretations to obtain “reasonable suspicion” of human misbehaviour, we do not yet know how to deal with this issue for misbehaviour of drones. As such, we may see anomalies, but we cannot infer whether a dangerous misbehaviour is in action. This creates either too many false positives or over-trusting aerial platforms. Therefore, there is still a gap in uncovering the causal motivations behind autonomous decision making and how these motivations can define the intention of a potential misbehaviour. In this talk, we aim to provide some of the recent developments and research conducted at Cranfield University to answer the scientific challenge of how we observe drones and interpret whether a real malicious activity is in action.

Short bio:
Dr. Adolfo Perrusquia is a Lecturer in Reinforcement Learning at Cranfield University. He received the MSc and PhD degrees both in Automatic Control from the Automatic Control Department at the CINVESTAV-IPN in 2016 and 2020, respectively. He holds a RAEng UK-IC Postdoctoral Research Fellowship (Oct 2021- Sep 2023) focused on predicting the intention of drones from observational data. He is a member of the IEEE and the IEEE Computational Intelligence Society, where he is currently chair of the reinforcement learning for robots task force of the ADP/RL committee. He is an Associate Editor of the IEEE TNNLS and a member of the EPSRC Peer Review College. His main research of interest focuses on reinforcement learning, data-driven control, inverse learning, machine learning and system identification.