Biomimetic robots: between promises and obstacles
Fly in swarms like bees, walk on walls like a gecko, be nimble like an octopus … Animals fascinate researchers who are inspired by them to create more efficient robots. Decryption.
“Nature has found very diverse and highly optimized solutions, after millennia of evolution, to an incalculable number of problems, notes Antoine Cully, Director of Adaptive & Intelligent Robotic Lab at Imperial College London. For example, how to walk, swim, climb, generate energy, how to deal with extreme conditions, etc. “. Problems that robotics engineers also face, drawing inspiration from nature’s adaptability. Get inspired, not copy.
“When you want to copy the organic, a scientific process that has a lot of meaning for research, you talk about biomimicry, explains Jean-Baptiste Mouret from the Inria Larsen team. However, when you take a piece of biology without going as far as copying the whole, it is bioinspiration. Sometimes it is very interesting to copy nature to find out how far science can go to better understand nature. And other times, you just want to be inspired by a particular phenomenon “. This is particularly the case with robotics developments.
Some brakes on a nanometric scale
Research in bioinspired robotics is numerous. Antoine Cully takes the example of the drone : “ An EPFL team is exploring new forms of wings to allow drones to fly for longer. In the same direction, researchers at Stanford are studying wings created from real feathers to improve the maneuverability of robots ”. Other research focuses on the dexterity of octopuses, the “ suction cups ” legs of geckos, the collaboration of ants, not to mention human hands… The roboticist Festo has also revealed last August his “BionicSoftHand 2.0”, a collaborative robot with one hand. “To copy a human hand, you would need a lot of motors, explains Jean-Baptiste Mouret. Thus, we study which movements are most useful for humans in order to reproduce them on the robot. We simplify the biological model. “
The brain and an individual’s ability to navigate in space are part of robotics research. “In drones, there are many applications around bioGPS, explains Agnès Guillot, doctor in psychophysiology and biomathematics, specialist in bio-inspired robotics. That is, how a drone can orient itself on its own to map its environment. In computational neuroscience, neurophysiologists have studied the nervous circuits to make a robot work ”. This bioGPS was for example used at AntBot, an ant robot capable of navigating outdoors, autonomously, without integrated GPS.
Without forgetting neuromorphic artificial intelligence, to realize real-time decision-making algorithms. A CNRS research team led by researcher Julie Grollier even reproduced artificial neurons and synapses in “ hard ”.
“Scientists are successful to reproduce a certain number of things, but not yet to imitate nature on a nanometric scale, affirms Agnès Guillot. And it is on this scale that progress is made. For example, we have never succeeded in building reliable artificial muscle. It is very complicated ”. The robots are rather equipped with motors – hydraulic at Boston Dynamics for example – and not artificial muscles. “ Boston Dynamics’ will is to succeed in achieving animal agility on machines , specifies Jean-Baptiste Mouret. They started with the paws with Big Dog but mostly did a lot of engineering that has nothing to do with biology. “
When research leaves the lab
Research in bioinspired robotics is of interest to industry. At the start of 2020, Spot robots, for example, took their first steps in offshore installations. NASA was inspired by the legs of the Gecko to develop Lemur (for Limbed Excursion Mechanical Utility Robot), a robot capable of climbing the outer wall of the International Space Station.
The study of optical flow – the speed of movement of objects on the retina – of insects has enabled the Institute of Movement Sciences in Marseille to develop sensors integrated today on drones. “By studying certain neurons in the eyes of insects, the researchers analyzed the strategy used by these animals to fly in very complicated environments” , specifies Jean-Baptiste Mouret.
“Despite all the recent advances in robotics in recent decades, it is still difficult to see a great diversity of robots operating outside the highly controlled environments of factories, with the exception of robot vacuum cleaners, adds Antoine Cully. My opinion is that designing this kind of robot requires large investments in order to ensure that the robots can cope with a wide variety of situations. ”
Photo de Une : Researchers are interested in the adhesive capacity of geckos legs.