Dr. David Bossens, Research Fellow @TAS Hub @UoS
email: d(dot)m(dot)bossens(at)soton(dot)ac(dot)uk

Welcome to the homepage of Dr. David Bossens, post-doctoral Research Fellow employed at the University of Southampton. I am currently involved in the UKRI TAS-Hub, a multi-institute organisation which aims at cross-disciplinary insights to the problem of safe and trustworthy autonomous systems. Throughout my career, my work has looked at designing AI systems with an open mind to the various sub-disciplines. My current research interests are:

1. providing improved safety to reinforcement learning systems;

2. solving the challenges associated with changing environments;

3. understanding different use cases, risks, and interventions for safe AI; and

4. drones on construction sites.

News

  • 01/2023: Presentation at the CFAR Rising Star Lecture Series
  • 09/2022: Start project on Safety and Desirability Constraints for Drones on Construction sites
  • 07/2022: First Safe RL Workshop at IJCAI:
    https://sites.google.com/view/safe-rl-2022
  • 06/2022: Paper proposing model-based robust-constrained method for safe exploration published at Machine Learning:
    Bossens, D. M., & Bishop, N. (2022). Explicit Explore, Exploit, or Escape (E4): near-optimal safety-constrained reinforcement learning in polynomial time. Machine Learninghttps://doi.org/10.1007/s10994-022-06201-z
  • 06/2022: Review paper on resilient robot teams published at Current Robotics Reports:
    Bossens, D.M., Ramchurn, S. & Tarapore, D. Resilient Robot Teams: a Review Integrating Decentralised Control, Change-Detection, and Learning. Current Robotics Reports 3, 85–95 (2022). https://doi.org/10.1007/s43154-022-00079-4
  • 02/2022: Paper for quality-diversity meta-evolution published at IEEE Transactions on Evolutionary Computation:
    Bossens, D. M., & Tarapore, D. (2022). Quality-Diversity Meta-Evolution: customising behaviour spaces to a meta-objective. IEEE Transactions on Evolutionary Computationc, 1–11. https://doi.org/10.1109/TEVC.2022.3152384