Mickey is a Robotics and Autonomous systems PhD researcher at the Bristol Robotics Laboratory (BRL) and the University of Bristol. After graduating with a first from Joint Mathematics and Computer Science at Imperial College London, he joined the FARSCOPE CDT program supervised by Professor Arthur Richards and sponsored by the Toshiba Bristol Research and Innovation Laboratory. His research focuses on optimal multi-UAV path planning for building inspection, in particular how guarantees can be provided despite vehicle failures. Most recently he has been developing a portable development and deployment infrastructure for multi-UAV experimentation for the BRL Flight Arena inspired by advances in cloud computing
Download my resumé.
FARSCOPE PhD in Robotics and Autonomous Systems, 2023
Bristol Robotics Laboratory, University of Bristol, University of the West of England, UK
MEng Joint Mathematics and Computer Science, 2018
Imperial College, London, UK
My research is focused on investigating the practical use of multi-drone and multi-agent systems. In particular, how the whole system can remain robust and reliable in the prescence of failures. The specific focus is on developing failure-aware multi-drone algorithms for coverage path planning for use in structural and infrastructure inspection. This work includes the following:
Supervised by Professor Arthur Richards (Bristol) and Professor Mahesh Sooriyabandara (Tohisba). Funded by ESPRC FASRCOPE Center for Doctoral Training in robotics and autonomous systems.
Responsible for providing teaching and support of the MSc Aerial Robotics MSc Programme.
Responsibile for the development and testing of a novel motion planning algorithm on a real drone. I worked closely with Dr Saurabh Upadhyay on re-implementing his fast motion planner for UAVs. I extended his algorithm to a receeding horizon formulation to enable dynamic reaction and obstacle avoidance.
Developed and publically released an educational game submission for the UK-RAS 2020 Robotics Week. I lead a team of 7 volunteer postgraduate students through the planning and development of the RoC-Ex: Robotics Cave Explorer game. The game is designed to teach children and teenagers how a robot senses the world and its environment through the exploration of a cave.
As the voluntary project manager, I suggested and pushed for methods to ensure that we would complete in time. This included organising 2 weekend “hackathon sessions” in which we all sat together to develop the game. As a developer, my responsibility was to integrate the models and missions into the Godot game engine, and to setup deployment and delivery of the game.
Tasked to explore and analyse a large distributed dataset. I introduced the use of cluster computing with Apache Spark and build analysis tools with Spark an Pandas for data analysis
I then implemented a novel recommendation engine model known as Collaborative Deep Learning in Distributed Tensorflow.
A major challenge of UAV research is facilitating the local development, deployment and testing of multi-UAV systems. Inspired by cloud computing, this work proposes Starling, a fullstack, compositional, containerised UAV infrastructure utilising ROS2, Gazebo, PX4, Docker and Kubernetes. By modelling individual UAVs as nodes in a compute cluster, our architecture is natively scalable, fault tolerant and allows for the flexible deployment of custom applications in both simulation and in reality. These technologies allow us to facilitate reproducible research while provide a lower barrier of entry for potential users, as well as the reuse of flight hardware for multiple projects. A multi-UAV path planning case study is presented to demonstrate the streamlined workflow of developing a controller from simulation to reality.
Graceful degradation is a potential advantage of Multi-Robot Systems over Single-Robot Systems. In aerial robotics applications, such as infrastructure inspection, this trait is desirable as it would improve mission reliability despite the use of failure-prone low-cost drones. The Reliability-Aware Multi-Agent Coverage Path Planning (RA-MCPP) problem finds path plans for each robot to maximise the probability of mission completion by a given deadline. This paper proposes a path planner for RA-MCPP formulated in continuous time, enabling more complex realistic environments to be considered. The proposed method (i) extends a reliability evaluation framework to evaluate the Probability of Completion metric on asynchronous strategies on non-unit lattice graph environments, and (ii) introduces a greedy-genetic meta-heuristic optimisation method as a scalable and accurate RA-MCPP solver. This method is shown to provide plans with higher reliability when compared with existing approaches in three real inspection scenarios.