A2I2 Extension: V&V of a semi-autonomous system in an operational environment

The A2I2 project have been developing untethered, underwater robotic platforms which make unmanned operations the standard for inspection and light intervention tasks for physical infrastructure, primarily in the offshore and nuclear domains. We are developing small, hover-capable Autonomous Underwater Vehicles (AUVs) equipped with a novel visual mapping system and enhanced & verified on-board autonomy. The designs of the platforms are scalable enabling work across domains. They are capable of conducting very close inspection and intervention tasks, such as cathodic protection surveys, coring, visual inspection and metrology and moving small items (e.g. in nuclear ponds).

The primary sensing apparatus is Rovco's SubSLAM stereo camera system, with associated computer vision software to allow the creation of 3D maps of the environment. The 3D data serves multiple purposes: enabling safe navigation in complex, cluttered environments by providing a 3D occupancy grid for guidance software; allowing localisation of targets detected by machine learning object recognition; and as metric survey output.

To enable the A2I2 robots to manoeuvre safely near sensitive infrastructure, the consortia are addressing current technological limitations:

* Perception in cluttered underwater environments
* Robust (Fail Safe) operation near sensitive infrastructure
* Precision manoeuvring and control near infrastructure
* Communication with robots in challenging underwater environments.

A2I2 provides a step-change beyond SotA in artificial intelligence control and communications.

The consortia members working in the offshore domain (Rovco, NOC and D-RisQ), supported by the Offshore Renewable Energy Catapult, will use this additional funding to demonstrate, verify and validate the system in the offshore operational environment. This will see improvements made to the perception system and the vehicle and would incorporate last response engine software to manage behaviour in the event of connection loss or other critical events. We will develop a safety case for the system and the project will culminate by demonstrating a semi-autonomous survey of an asset (i.e export cable and monopile) with a human in the loop. This is a significant step-change in operational capability. The system will ultimately remove the need for pilot operators to be present on the vessel, as the vehicle requires only supervision which can be conducted remotely. This in turn reduces the costs and environmental impact of operations by allowing smaller, less fuel consuming vessels to be used. The current drive to reduce crew as a direct impact of COVID-19 is accelerating the adoption of remote and autonomous technologies, this follow-on project performs the V&V demonstrations to accelerate the technology to market.