When western shovel-nosed snakes slither swiftly through mixed terrain, they respond passively to obstacles without neural input. That is the surprising conclusion of Daniel Goldman and colleagues at the Georgia Institute of Technology in the US. What is more, the paths taken by the reptiles bear more than a passing resemblance to the trajectories of quantum particles. The team believes that the snake’s mechanical response to obstacles could inspire new types of limbless robots that could traverse complex environments.
“Modern robots are fantastic if the environment is really simple, but when complex, like in the real terrestrial world with branches, leaf litter and sand, then robot performance degrades very quickly,” explains, Goldman who heads up the Complex Rheology And Biomechanics (CRAB) lab.
Snakes are capable of slithering through and over an amazing variety of terrain, making them particularly interesting animals to mimic. “If we can incorporate some of the principles by which living systems control limbless bodies, it could one day give robots the ability to go anywhere in the terrestrial world,” said Goldman.
Do the locomotion
In previous work, CRAB lab’s Jennifer Rieser discovered that when a snake-like robot (moving with a simple wave pattern) collided with an obstacle it interacted persistently until it pushed past and moved off at an angle to its initial trajectory. Rieser measured a predictable scattering pattern, and then teamed-up with Perrin Schiebel to investigate whether the same locomotion dynamics were found in real snakes.
Limbless animals have complex dynamics – they can bend and twist in all directions to take a myriad of shapes – and understanding this complexity can be very difficult. So Rieser and Schiebel simplified matters by focussing on the western shovel-nosed snake, which uses relatively simple head-to-tail waves to slink swiftly over the desert sands at night.
Using eight blindfolded snakes, they placed one creature at a time into a carpeted arena with a 1D lattice of dowel posts (see figure). A high-speed camera was used to track the snakes as they performed 253 manoeuvres through the lattice.
“When we put the snakes down in the arena, they started moving using the same waveform they use on desert sand,” explains Schiebel. “They would then encounter the dowel grating, pass through it, and continue on the other side still using that waveform.”
Snake diffraction
The first and most unusual observation the physicists made was that the overall pattern of how the snakes emerged from the lattice was very similar to how photons interact with a diffraction grating.
Goldman was surprised to see this and explains, “In the quantum world you see individual particles interacting with a regular array and emerging in preferred angles, but I didn’t expect to see such phenomena when working on animals and robot”.
Biomechanics expert Noah Cowan at Johns Hopkins University told Physics World that this connection is “striking” adding, “It has a remarkably similar diffraction pattern to a photon diffraction grating”. Indeed, Goldman is keen to discover whether the diffraction of sub-atomic particles can be mimicked and investigated using snakes.
Passive dynamics
Although the live snakes move much faster through the lattice than Rieser’s robots, their scattering patterns are very similar. This suggests that the navigation-control system used by the snakes resembles that used by the robots – an “open-loop system” that lacks sensory feedback and response.
“When the desert snakes scatter they are moving so fast we don’t think the nervous system can correct for the perturbations,” said Goldman. “And based on the diffraction pattern and force pattern generated on the array, the snakes appear to maintain this motor programme and close a feedback loop when they are in the lattice.”
To further corroborate this open-loop, purely passive mechanical response, the CRAB lab team modelled geometric snake muscle dynamics with open-loop controls, and found it to closely match the experimental snake scattering pattern.
Muscular Robotics?
Biorobotics specialist Auke Ijspeert of the Swiss Federal Institute of Technology at Lausanne, praised the teams application of physics to robotics. “The core message is that when you design a robot it’s important to think about imbedding these interesting biomechanical features,” said Ijspeert.
Reptile skin inspires super-slippery steel surfaces
Cowan agrees that applying these features could help a limbless robot interact more robustly with complex environments. He comments, “I think it’s another signpost underscoring that locomotor ability is not going to be solved by artificial intelligence alone.”
The CRAB lab members are keen to test this concept by applying snake-like muscles to their robot and testing it within a complex environment.
The research is described in Proceedings of the National Academy of Sciences.
