The return to an aquatic life, the evolution of propulsive efficiency & biomimetics – DRAGON2

The resolutely multidisciplinary Dragon-2 project was based on interactions between different disciplines. Zoology, ecology: to fill in the gaps in our knowledge of how different species of snakes swim, we developed large swimming benches equipped with high-speed cameras. In practice, we used water drainage pipes cut lengthwise. The pieces were placed end to end. Relatively easy to manufacture, dismantle and transport, these swimming tanks enabled us to acquire the richest set of data ever collected. The diversity of the kinematics, ecologies and morphologies of the snakes studied provides a very valuable basis for understanding how natural selection has shaped and optimised snake swimming. Fluid mechanics: as it is impossible to measure the energy consumption of swimming snakes with the necessary precision, i.e. the expenditure associated with each lateral undulation, the idea was based on the use of indirect techniques capable of estimating the kinetic energy ‘left’ by snakes in the environment in the form of eddies. To visualise and measure the vortices, 3D particle image velocimetry was used, which is difficult to implement and had never been used to study snake swimming before. A long development phase was necessary. Microscopy: the structures and microstructures present on the surface of snake scales may play a hydrodynamic role. To verify this, scanning electron microscopy (SEM) in variable pressure mode (without prior metallisation), stereo profilometry and 3D opto-digital microscopy were used to study the scales of numerous snake species from the collections of the organisations involved in the project. Mechatronics and Robotics: in order to design and build a robot that swims like a real snake and can be used as an instrument to experimentally test hypotheses on the hydrodynamics and energy efficiency of swimming, the skeleton of snakes was studied. Stiffness and resistance to imposed bending tests were carried out using a device developed specifically for this purpose. The information obtained was used to calculate the shape of the artificial skeleton, composed of bio-inspired vertebrae connected by gimbal-type joints and stops, and the types of materials to be used to reproduce the behaviour of a snake's undulating body. Numerical modelling and simulation: these were carried out mainly using Navier-Stokes equations to construct the basic models. An immersed boundary and volume penalisation approach was used to represent the fact that the body of the swimming snake deforms during swimming.

Thanks to fieldwork carried out in various countries, the diversity of snakes' swimming patterns has been linked to the diversity of their lifestyles. The most specialised species, which are strictly arboreal, burrowing or aquatic, use specific and distinct swimming kinematics. Species with less specialised lifestyles, such as terrestrial, semi-arboreal, semi-burrowing or semi-aquatic species, differ from one another on average, but their kinematics are largely overlapping. Specialisation towards the aquatic environment has not led to an increase in swimming speed, but rather to an increase in energy efficiency. This result is surprising, as in all other terrestrial vertebrates that have adapted to aquatic life, particularly birds and marine mammals, swimming speed is much higher than in the most closely related terrestrial species. Snakes have followed a different evolutionary path. Hydrodynamic measurements taken in the laboratory have provided full support to the theoretical predictions about the shape and behaviour of the vortices produced by the lateral undulations of snakes when swimming. Beyond this technical feat, it has been possible to estimate the energy expenditure required by snakes to generate water movement. Snakes appear to be particularly agile at moving through water, creating few vprtices, efficiently and probably very economically. The diversity of the microstructures of snake scales could not be linked to adaptation to aquatic life, which is quite logical given that swimming speed does not appear to be a major criterion. The highly bio-inspired approach to designing the snake robot has been successful. From the early stages of vertebrae creation to the 3D printing of the snake, which is over a metre long, trial and error led to a robot snake with fluid and economical movements. A single 9-volt battery allows it to swim for 4 hours, which no competing robot is capable of doing. The modelling and digital simulations required significant computing power because the processes involved are complex. But above all, they produced results consistent with those obtained in other parts of the project. For example, the snake's body is the main source of propulsion, whereas in most fish the tail is the main propulsive organ. This result once again highlights the unique characteristics of snake swimming.

The project focused on understanding how the snakes studied swim in a specific situation: a straight line. The complexity of the phenomena studied required limiting the number of parameters and their interactions. The snake robot was designed and developed with this simplification in mind. Real snakes almost never swim in a straight line. Instead, they change course and perform complex manoeuvres, for example to capture prey in the water. A snake robot should also be able to move in three dimensions in water in order to be useful, both as an instrument for testing hypotheses generated from observations of live snakes and as a tool for exploring the environment or inspecting technical installations. Based on new knowledge, the natural prospects for the project are to study swimming in three dimensions, but also during transitions between environments, for example when a snake enters or leaves the water. An approach broadly equivalent to that adopted in this project could be applied. That is, studying the complex kinematics of snakes swimming as they change direction, surface or dive, or even reverse direction, as some are capable of doing. Ideally, these measurements could be made by studying snakes in the wild and under controlled conditions using swimming tanks that allow individuals to be filmed from different angles. Measurements of the vortices created by snakes changing direction have been taken in the laboratory during the final year of the project,, showing that this type of approach is realistic. At the same time, the snake robot will need to draw on information obtained from live snakes to configure its movements. Similarly, digital models will be able to draw on observations made in the field and in the laboratory. In short, the outlook is to continue using an approach based on observing the extraordinary diversity of snakes and bio-inspiration to test hypotheses and develop high-performance robots.

The return to an aquatic life has shaped many organisms and profoundly impacted fossil and extant marine ecosystems. Understanding how terrestrial organisms adapted to a drastically novel environment poses fundamental challenges, however. Indeed, in most lineages this major transition entailed deep alterations of locomotor modes that impede straightforward comparisons among evolutionary stages. Snakes constitute an exception. A single undulatory locomotor mode is efficient both on land and in water. Our key hypothesis is that the return to an aquatic life, frequently observed in snakes, was mediated by the optimization of the undulatory kinematics without a radical alteration of this locomotor mode. We postulate that optimized undulations minimize resistance and maximize propulsive drag. Energetic efficiency of swimming should thus be higher in aquatic species compared to closely related terrestrial species. Yet, quantifying swimming performance, undulatory kinematics and energetic expenditure simultaneously is technically challenging, especially using non-invasive techniques. Furthermore, physical effort can be partly decoupled from oxygen consumption in snakes, making classical techniques (e.g. respirometry) imprecise. One option is to measure the drag coefficient of swimming animals: the vortical structures produced at each time interval may be use to accurately quantify the efficiency of locomotion. Thus, fluid mechanics and numerical modelling are alternatively solutions allowing to tackle this complex problem involving deformable structures. This fundamental project based on biology and fluid mechanics also relies on robotics. Bio-inspired snake-robots will be designed to experimentally assess the relationship between kinematics, energy expenditure and hydrodynamic drag. This multidisciplinary project includes 5 work-packages. WP1: motion capture and 3D-kinematic analyses will be used to analyse the undulatory kinematics (frequency, amplitude) of swimming snakes in the laboratory. Drag will be measured using volumetric particle image velocimetry. A range of terrestrial, amphibious and aquatic species will be tested. WP-2: key parameters obtained in WP1 will be used to design swimming robots to test the influence of swimming kinematics on propulsive and resistive forces. WP-3: skin surface structure of a wide diversity of snakes will be examined using scanning electron microscopy, micro-CT scans and gel-based stereo-profilometry. 3-D reconstructions of skin surfaces will be tested in a flow tunnel to examine their tribological properties. WP-4: the information collected will provide the basis for numerical simulation analyses of the energetic efficiency of displacement. The objective is to develop a predictive model that integrates body size, body shape, skin structure, undulatory kinematics to obtain the energetic efficiency of any swimming snake (or robot). Ultimately, we plan to automatically extract and analyse undulatory kinematics from videos of swimming snakes to derive the cost of transport associated. WP-5. the predictive model will be used to estimate the swimming efficiency of a large number of species for which video records will be obtained in the field. The large and unique collection of kinematics representative of the extraordinary diversity exhibited by snakes will allow us to frame analyses into a phylogenetic context. Factors like body size, foraging mode, reproductive status, and sex will be implemented. This huge data set will also provide multi-optimization criteria for robot prototypes. The numerical codes developed during this project and the data bases will be registered (INPI). Besides fundamental objectives in evolutionary biology, this project based on state-of-the-art techniques and measurements on living snakes to understand the hydrodynamic efficiency of undulatory swimming represents a unique opportunity for French laboratories to participate in the race to develop snake-robots.