Aquatic Agents Ltd

Building automonmous soft-body underwater vehicles 

We develop and design underwater vehicles as platforms for research into novel propulsion methods inspired by animals such as whales, squid, turtles, rays, penguins and tuna. We also build models around these technologies such as fish navigating, migrating and schooling which have been used as part of the environmental impact assessment of aquatic projects, but which now can be used to inform robot development. We are a UK based company located near London.


Email :


Animals - Base Unit 1: Mobula

We're aiming to make underwater vehicles that use flapping propulsion. Animals are our inspiration. Turtles, sharks, penguins, and rays have all been developed over millions of years for efficiency, stealth and resilience. We have designed a number of base systems to test how efficient flapping propulsion is likely to be using modern materials. We explore the design parameters to discover the most efficient, powerful and practical solutions. The first base unit is a Mobula ray design. It is one of our design principles to base the designs on appropriately sized animals; size is especially important for hydrodynamic performance. The CAD drawings show the Mobula with a wingspan of about 1.5 m, followed by an image from the first pool test of the prototype. A potential use case for this type of vehicle would be coral reef surveying. Quiet, efficient, and low environmental impact.

Animated GIF



Models of multiple aquatic animals range from particle models where the animals are passive drifters at the mercy of current, through individual based models where the individual model animals have a set of methods to react to their environment, such as swimming, migrating or avoiding land. The final level of sophistication in models are agent based models. In agent based models each animal is modelled as an individual with agency which can react to the physical environment as before, but can also react to the other participants in the model. This interaction between model participants can result in emergent behaviour that was never explicitly coded in at the start. The canonical example are models of flocking or schooling, where realistic natural patterns of large schools and flocks can be replicated by simple rules enacted between the many individual model participants. We have put our own spin on these types of model by the addition of chaotic water currents to research the balance between school fidelity and the effort it takes. The patterns are caused by simple rules but the results are dynamic and interesting. Autonomous underwater vehicles need to act as independent agents and intercommunication will be the critical to their success.