The oral surface of sea stars (starfish) is lined with arrays of tube feet that enable them to achieve highly controlled locomotion on various terrains and to even gallop and bounce. The activity of the tube feet is orchestrated by a nerve net that is distributed throughout the body; there is no central brain. How such a decentralized nervous system produces a coordinated locomotion is yet to be understood. To examine the sensorimotor control underlying the sea star locomotory behavior, we developed mathematical models of the biomechanics and sensorimotor control of the tube feet. In these models, the feet are soft actuators that are coupled mechanically through their structural connection to the sea star body, and whose power and recovery strokes are dictated by local sensori-feedback control loops. We found that these minimally-coupled tube feet coordinate to generate robust forward locomotion, reminiscent of the crawling motion of sea stars. We also found that the sea star model can transition from crawling to bouncing, and it can robustly locomote on various terrains, and under various heterogeneity in the tube feet parameters and initial conditions, akin to experimental observations. These findings improve our understanding of the Echinoderms decentralized nervous system and could lead to novel designs and control rules for autonomous robotic systems.

Anisotropy is very often present in nature to support an associated mechanical function (softness in one direction, stiffness in another). Nowadays, we can reproduce a similar approach thanks to 3D printing and in the context of the use in soft robots, this opens a very large field of possibilities for design. But sometimes this field is so wide that we find ourselves helpless due to the lack of methodology.

During this talk, we will present a modeling approach allowing to calculate the kinematics of the robot taking into account the anisotropy and another work on very fast calculation methods by model reduction. We will also discuss some recent results on the control of soft robots and see how they can apply in this context.

Finally, we will try to draw perspectives to continue to build a coherent methodology that goes from the design to the use of these anisotropic soft robots.

Soft robots typically depend on tethered connections to external hardware for power and actuation. While fine for early-stage research or operation of immobile robotic systems, dependency on tethered hardware can impair the mobility of soft robots and limit their ability to approach the maneuverability of natural organisms. In this talk, I will review recent methods to cut the cord and achieve untethered functionality for a variety of applications in soft robotics, including limbed locomotion and swimming. In particular, I will focus on recent implementations that incorporate shape memory materials such as shape memory alloy and liquid crystal elastomer. I will also highlight ongoing efforts to create new computational tools for modeling the motion and surface interactions of untethered soft robots. Based on continuum mechanics, finite element analysis, and emerging techniques in computer graphics, these tools represent another critical requirement for soft robot autonomy by potentially enabling on-board computational intelligence and adaptive decision making. As time permits, I will also briefly touch on device-level subcomponents that will enable untethered functionality in future soft robotic systems. This includes elastically deformable batteries, energy harvesting transducers, and sensors that wirelessly transmit recordings of mechanical deformation.

Limbless robots have the potential to navigate environments impossible for more conventional legged robots. Many such devices operate with serially connected rotational motors which generate waves of bending traveling down the body. To date, most such devices move with difficulty and require significant active control to traverse heterogeneous terrain [e.g. Travers et al, RSS, 2016]. Here, to create a robophysical model that can capture the mechanical “diffraction” patterns observed in biological snakes transiting arrays of posts [Schiebel et al, PNAS 2019], we develop a new kind of limbless robot whose inherent mechanics and dynamics allow good locomotor performance in heterogeneous laboratory environments without active electronic feedback control. In the diffracting snake study, we discovered that passive body buckling, facilitated by unilateral muscle activation, allowed obstacle negotiation without additional control input. To model these features, we created a ~40 cm long limbless model with 8 joints and 16 motors. The joint angle is set by activating the motor on one side, spooling a cable around a pulley to pull the joint that direction. To model the snake muscle activation patterns [Jayne, J. Morph, 1988], we programmed the motors to be unilaterally active and propagate a sine wave down the body. When a motor is inactive, it is unspooled so that its wire cannot generate tension. Pairs of motors can thus resist forces which attempt to lengthen active wires but not those pushing them shorter, resulting in a kinematically soft robot that can be passively deformed by the surroundings. The robot can move on hard ground when drag anisotropy is large, achieved via wheels attached to the bottom of each segment, passively re-orient to track a wall upon a head-on collision, and traverse a multi-post array with open loop control facilitated by buckling and emergent reversal behaviors. Our design in which control is offloaded into the the mechanics of the robot echoes a successful strategy in legged robots [e.g. Saranli et al., IJRR, 2001], and could lead to limbless robophysical models for diverse living systems (including nematode works and snakes) as well as improved performance of limbless robots in real world terrain.

Non-linear force-displacement relationships offer interesting capabilities for soft robotics. We will explore examples for: (i) inflated structures with flexible but inextensible shells, (ii) soft cellular robots with strain-stiffening skins, and (iii) soft silicone fingers with embedded bistable springs.

The bodies of Soft Robots are designed so to present intelligent behaviors even when disconnected from the robotic brain. Effectively exploiting this intelligence requires a complete perspective change apropos the classic control of rigid robots. Rather than imposing a prescribed behavior detached from the natural behavior of the robot, we want the controller to work in synergy with the intelligent body.

For example, it is intuitively clear that the ability of robotic systems to perform efficient oscillatory motions can be massively enhanced by introducing elastic and soft elements in their design. Thus, Soft Robots should be especially suited to perform oscillatory tasks (e.g. locomotion, industrial repetitive tasks). Yet, formalizing this intuition and exploiting these capabilities came out to be all but easy. In this talk, I will discuss our recent efforts in generalizing modal analysis to Soft Robots (aka nonlinear mechanical systems subject to an elastic potential field) through the introduction of a nonlinear generalization of the linear Eigenspaces: the Eigenmanifolds. I will then discuss how this formalization can be combined with control theory towards the hyper-efficient and goal-directed oscillatory behaviors in soft robotic systems.

Due to their continuous and natural motion, soft robots have shown potential in a range of robotic applications including bio-inspired robots, wearable devices, VR/AR and industry. Despite these advantages and the rapid developments in recent years, robots using soft materials still have several challenging issues to be addressed. In this talk, I'll present some of our recent work in the area of soft robotics, covering the topics of optoelectronic haptic sensing for robots and soft haptic actuators for humans. The hidden key components behind these two functions are soft sensors and soft actuators. In our work, we have developed accurate, repeatable, and stretchable sensors using soft transparent materials, and high-power-density, long-lifetime, soft actuators using dielectric elastomers. Soft robotic technology not only enables robots to “touch and feel”, but also empowers human with “a feeling of being touched”.