Underwater Navigation & Robotics

Mounting pressure from multiple sectors in response to climate change and the need for increased conservation is dramatically altering the way that we interact with the oceans. There is a major push towards reducing vessel size and crew numbers at sea and to accelerate operations so that off-shore time is minimised. This is being driven by not only third-party clients of subsea services, but operators themselves to reduce costs and carbon emissions.

Emerging technologies for autonomous water-based vehicles are driving a new approach to off-shore surveys, inspection and data collection. Small-scale electrically powered AUV and UAV are delivering massive reductions in vessel costs and traditional diesel engine emissions through super-fast deployment, superior mobility and speed. The adoption of complex AI and high-performance navigation systems and other sensing technologies in conjunction with these vehicles is providing greater scope for subsea operations without the previous massive capital expenditure, required skill sets and logistics.

The benefits of autonomous and uncrewed water vehicles are huge. These vehicles require no equipment for human operators, which allows them to be incredibly small and very much bespoke to the intended tasks. Smaller, lighter and less complex equipment is easier to manufacture, uses fewer resources and is more affordable to operate and maintain.

Autonomous and semi-autonomous vehicles are not only reducing capital expenditure in subsea services, but revolutionising it. The future includes, for example, robotic vehicles in the offshore field that can autonomously make informed AI based decisions. The effects are noticeable drops in carbon emissions, significant improvements in efficiency, and reduced time-to-mission. These heavily reduce costs for crew, vessels and ancillaries, and provide the ability to more frequently carry out survey and data acquisition missions.

Advanced Navigation has long advocated the use of artificial intelligence and deep learning artificial neural networks. This technology is a feature that enables navigation systems, for example, to use previous data to identify and compensate for sensor drift, causing accuracy to improve, over time. Continued refinement and adoption of AI bring benefits to automated systems and robotics. Using AI to interconnect sensor suites and vehicle subsystems enables superior control and the ability for the vehicle to react to its surroundings without human intervention. For example, collision avoidance and performing actions based on recognition of an object of type.

AI is well suited to fusing asynchronous data from different sensors and using this data to make cohesive estimates of position and motion. That is, integrating and fusing data that arrive at different times from sensors that update at different speeds; for example, 1000 Hz, 1 Hz, or in the case of USBL, varying speed depending on acoustic travel time.

Hydrus is an example of AI that deeply fuses multiple subsystems. As a scenario, Hydrus can be on a mission to capture images of coral, for example, and during the capture of an image, a sudden shift in current may cause the vehicle to change position. Through AI, the image is analysed and a decision made as to whether the quality is acceptable. If not, the robot returns to the original capture position, checks conditions such as lighting, turbidity and focal length, retakes the image then continues the mission.

USBL² uses a transducer on both the subsea asset (ROV, AUV) and the surface vessel. This enables using a shared algorithm and performing USBL calculations at both ends, which produces significant gains in accuracy and reduces measurement uncertainty. Each Subsonus transducer has an INS, which enables full orientation and heading to be calculated and resolved between both the vehicle and surface vessel. USBL² provides improved accuracy and levels of performance with absolute heading and position being transferred acoustically between the transducers.

USBL² is a highly effective and complication-free solution for applications that require very high acoustic navigation performance. USBL² provides reliable absolute position and heading for assets via acoustically transferred GNSS data between the Subsonus units. Obtaining absolute subsea heading data in this way is extremely cost-effective when compared to more complex and expensive alternatives, such as fibre-optic gyroscope (FOG), and is also immune to magnetic interference.

An ideal solution is to adopt autonomous underwater robotic drones that can remain resident to the inspection area and can repeat missions without intervention. Hydrus is such a system. It can be programmed as a mission that it repeats at predefined intervals; for example, inspecting a port seabed for shipping lane obstructions or for dredging. On mission completion, the captured data is uploaded automatically while the Hydrus docks and recharges.

Perhaps more critically, Hydrus can carry payload software that can leverage the robot AI and enable it to capture very specific information. For example, to identify corrosion in equipment moorings, cracking or movement of concrete footings, dam wall inspections, detect objects of type (shape and size), and monitoring changes in marine habitat, such as water composition and coral growth or decline.

Using such a system can very significantly reduce costs for not only capital expenditure but in crewing, divers, insurance, consumables (vessels and crew) and time to mission.

There will typically be several navigation systems and technologies used in an AUV navigation system. For example, a means of establishing position may be required. For absolute position, this often means a USBL with GNSS input that can transmit bearing data to the AUV. An INS may be used in continued navigation, in a dead-reckoning sense, to estimate position based on its inertial sensor inputs, with reference to the last known absolute position or previous estimated position. Acoustics systems, such as sonar and DVL (Doppler velocity log), can be used to improve positioning accuracy by providing additional motion data based on the Doppler shift in sound signals bounced off the seabed. This data is another input to the INS signal processing that introduces additional velocity information, the speed and direction of currents, and vehicle orientation and heading.

Depth in open water is often determined by pressure sensors. Input data from the sensors are fed into an algorithm that applies filtering of the data and introduces additional variables, such as water temperature and salinity, and INS data to provide greater accuracy. For altitude from the seabed, acoustic DVL may be used. This system emits acoustic signals that bounce off objects. The returned signals are processed and, using time over distance and velocity of sound in water, can determine distance from the seabed.