Efficient Soft Robots with Embodied Intelligence

Soft robotics is largely motivated by Embodied Intelligence, i.e. the behaviour emergent from external interactions. According to this well-known paradigm, behaviour is not only controlled by computation but emerges from the physical interaction of the body with the environment [1]. It then depends on the mechanical properties of the body itself, on its morphology, on the environment, on the tasks at hand. This concept of embodied intelligence brought to rethink bodyware in robotics, compliance became key and either compliant joints or soft materials have been used.<br/>Using embodied intelligence means increasing efficiency of movements, also in energetic sense. An extreme example of embodied intelligence is the passive walker [2], a completely passive walking robot. Its two legs can passively produce sequential steps on a slope, where gravity can perturb the system balance. Of course, this is an extreme example, working in a narrow ecological niche, i.e. an inclined plane. However, few works show that the ecological niche can be extended to flat terrains, by adding a limited number of active movements [3], simple to control and energetically cheap, without losing the advantages of embodied intelligence. Full advantage is still taken of the effect of gravity, as part of the leg movement derives from gravity force. As well, energy consumption is greatly reduced, thanks to this use of embodied intelligence.<br/>In robotics, computational efficiency is instrumental to improve robot behaviour, today hindered by complex and slow control systems. Elegant, computationally frugal solutions that we find in nature [4], like embodied intelligence, are rarely adopted in robotics, where brute-force approaches are often put in place. Computational efficiency is key for reducing energy need, one of the hindering factors for the wide spreading of robotics. Low energy needs can be fulfilled by energy amounts available in the environment, in food, air and water motion, or light. Computationally and mechanically efficient robots, endowed with energy recovery mechanisms, can tend to become energetically autonomous.<br/>Bioinspired soft robots leverage on efficient mechanisms learnt from living beings and find application in explorations, among others. They can access remote areas, confined spaces, complex or collapsed structures, both in land and at sea. Marine applications of soft robots give interesting challenges for bioinspired swimming, locomotion and manipulation underwater. Principles from octopuses and crabs enabled the development of efficient legged robots walking underwater with emergent self-stabilizing gaits [5], with low energy consumption and zero-power station keeping. Likewise, octopuses show unconventional grasping and manipulation movements, requiring soft arms, extremely efficient in water.<br/>Overall, soft robots support a vision for future robots where they can become autonomous, not only in their sensory-motor behaviour, but also in growing, learning, adapting, self-healing and powering themselves [6]. This vision for life-like robots provides challenges and opportunities for exploring materials, power sources, sensors, actuators and mechanics for untethered soft robots.<br/>[1] Pfeifer R, Lungarella M, Iida F 2007 Self-organization, embodiment, and biologically inspired robotics <i>Science</i> 318(1088)<br/>[2] McGeer T 1990 Passive Dynamic Walking <i>The International Journal of Robotics Research</i> 9(2) pp.62-82<br/>[3] Collins S H, Wisse M, Ruina A 2001 A 3-D Passive Dynamic Walking Robot with Two Legs and Knees <i>The International Journal of Robotics Research</i> 20 (7) pp.607–615<br/>[4] Berthoz A 2012 <i>Simplexity: Simplifying principles for a Complex World</i> (Yale University Press)<br/>[5] Picardi G, Chellapurath M, Iacoponi S, Stefanni S, Laschi C, Calisti M 2020 Bioinspired underwater legged robot for seabed exploration with low environmental disturbance <i>Science Robotics </i>5(42)<br/>[6] Mazzolai B, Laschi C 2020 A vision for future bioinspired and biohybrid robots <i>Science Robotics</i> 5(1)