Skip to content
Research Fellows Directory

Yvan Petillot

Professor Yvan Petillot

Research Fellow

Organisation

Heriot-Watt University

Research summary

In this research, we are exploring the key technologies required to perform autonomous manipulation (i.e. without human intervention) subsea. Why is this important? It matters because humans can only dive to limited depth and unmanned platforms such as remotely operated vehicle (the subsea version of a remote controlled boat) require skilled operators and big and expensive support ships. It also matters because we need more than ever to explore, exploit and preserve remote maritime environments, from the arctic to the deep sea. The key barriers to perform underwater manipulation are in the control (how to move the platforms and their arms around) and the computer vision (how to sense and understand the scene using algorithms) arenas. This is where the core of our research lies. We have been developing techniques to control vehicles equipped with manipulators in the presence of disturbances such as currents and waves. On the control side, this involves understanding the complex dynamic interactions between the vehicle and its arms and the environment and developing schemes that can adapt in real time to the changes of the environment. On the computer vision side, it involves developing techniques to automatically interpret scenes, detect objects of interest and direct the system to manipulate these objects using visual feedback. The applications are numerous, from back box search and recovery to environmental monitoring and repair. The offshore industry would also benefit from systems that can be deployed remotely over long time periods to perform periodic inspection and maintenance of subsea assets without the need for support vessels. This is particularly true for harsh or dangerous environments.

Interests and expertise (Subject groups)

Grants awarded

Advanced autonomy in the subsea domain

Scheme: Industry Fellowship

Dates: May 2013 - Apr 2015

Value: £109,460