3D underwater eel robot

Compared with our technological achievements, the performance of fish is nothing short of breathtaking. Among these are their prodigious acceleration capacities, up to twenty times gravity, their speed in excess of 70 km/h, their extraordinary manoeuvrability: they can turn through 180° without slowing down and over radii of curvature of the order of a tenth of their length, whereas today's vehicles have to slow down by half and take radii of curvature of the order of ten times their length. In terms of efficiency, their output is ten times greater than that of our best submarines, and so on. These figures alone motivate current efforts to understand and reproduce fish solutions on our robotic systems. In this field of biomimetics, the first of the difficulties encountered is described in the following terms:

"Replicating the performance of a fish by simply imitating its form and function would be impossible, because developing a vehicle that flexes smoothly and continuously is beyond the current capabilities of robotics" .

For this reason, the continuous nature of fish is the main difficulty for research in this field. The aim of this project is to reinforce biomimicry by creating a prototype eel robot that is "more continuous" than its current counterparts. To this end, the prototype's mechanical architecture is based on the serial stacking of parallel platforms sheathed by a continuous flexible member playing the role of a skin. The modeling itself is based on dynamic models known as macrocontinuous (macro for macroscopic), based on the theory of continuously-actuated Cosserat beams.

To achieve this objective, we began the project with a biomechanical study. On the basis of this study, the assisted design of a continuous macroscopic model (macrocontinue) was launched and carried out in parallel with a polyarticulated model more faithful to the technological reality of the future prototype. Finally, from the outset, modeling of the contact between the fluid and the structure was initiated. From the outset, we adopted a hierarchical modeling approach for both robot mechanics and fluid-structure contact. Based on these models and associated simulators, the control system is currently being studied and will eventually be implemented on a computer architecture.