Mark’s research is focused on the Flying Machine Arena, specifically on selected topics for increasing autonomy of the system. Mark’s research can be broken down into the below sections.
Trajectory generation for quadrocopters
The problem to be solved here is that of finding the inputs which will take a quadrocopter from its current state, to a desired end state, in a given amount of time. This occurs, for example, if a quadrocopter is tasked with hitting a ball towards a target, or has to land on a moving platform. A key requirement is that these trajectories must be solvable in real time, so that they can be used in feedback by the system to correct for disturbances and adapt to changing situations. See it in action!
Fault tolerance and failure mitigation
The Flying Machine Arena is a large and complex system, but one that we would like to be able to confidently show people. One of the resulting challenges is making sure that the system reacts well to failures and faults. Examples of such faults include a failure of the global sensing system, failure of the radio communication system or actuator failure on the vehicles. See it in action here, and here.
Localisation and state estimation
Flying vehicles have particularly interesting dynamic properties, and present opportunities for new state estimation algorithms and approaches. Furthermore, technologies such as ultra-wideband ranging are becoming more and more available, allowing the creation of new sensing strategies allowing a flying vehicle to estimate its dynamic state and thus fly.
Typical multicopters have an even number of propellers, appearing in pairs of counter-rotating propellers. These designs have the advantage that the input torques can be made to balance, while the vehicle is hovering. Alternative designs, where the input torques cannot be made to balance, offer interesting dynamics and a new take on multicopters. An example of such a vehicle, which uses only two propellers, can be seen here.
The full list of publications is available here.