Softworms: the design and control of non-pneumatic, 3D-printed, deformable robots

Robots that can easily interact with humans and move through natural environments are becoming increasingly essential as assistive devices in the home, office and hospital. These machines need to be safe, effective, and easy to control. One strategy towards accomplishing these goals is to build the robots using soft and flexible materials to make them much more approachable and less likely to damage their environment. A major challenge is that comparatively little is known about how best to design, fabricate and control deformable machines. Here we describe the design, fabrication and control of a novel soft robotic platform (Softworms) as a modular device for research, education and public outreach. These robots are inspired by recent neuromechanical studies of crawling and climbing by larval moths and butterflies (Lepidoptera, caterpillars). Unlike most soft robots currently under development, the Softworms do not rely on pneumatic or fluidic actuators but are electrically powered and actuated using either shape-memory alloy microcoils or motor tendons, and they can be modified to accept other muscle-like actuators such as electroactive polymers. The technology is extremely versatile, and different designs can be quickly and cheaply fabricated by casting elastomeric polymers or by direct 3D printing. Softworms can crawl, inch or roll, and they are steerable and even climb steep inclines. Softworms can be made in any shape but here we describe modular and monolithic designs requiring little assembly. These modules can be combined to make multi-limbed devices. We also describe two approaches for controlling such highly deformable structures using either model-free state transition-reward matrices or distributed, mechanically coupled oscillators. In addition to their value as a research platform, these robots can be developed for use in environmental, medical and space applications where cheap, lightweight and shape-changing deformable robots will provide new performance capabilities.

Energetic analysis and experiments of earthworm-like locomotion with compliant surfaces

The energy consumption of worm robots is composed of three parts: heat losses in the motors, internal friction losses of the worm device and mechanical energy locomotion requirements which we refer to as the cost of transport (COT). The COT, which is the main focus of this paper, is composed of work against two types of external factors: (i) the resisting forces, such as weight, tether force, or fluid drag for robots navigating inside wet environments and (ii) sliding friction forces that may result from sliding either forward or backward. In a previous work, we determined the mechanical energy requirement of worm robot locomotion over compliant surfaces, independently of the efficiency of the worm device. Analytical results were obtained by summing up the external work done on the robot and alternatively, by integrating the actuator forces over the actuator motions. In this paper, we present experimental results for an earthworm robot fitted with compliant contacts and these are post-processed to estimate the energy expenditure of the device. The results show that due to compliance, the COT of our device is increased by up to four-fold compared to theoretical predictions for rigid-contact worm-like locomotion.

Phase coordination and phase-velocity relationship in metameric robot locomotion

This research proposes a new approach for the control of metameric robot locomotion via phase coordination. Unlike previous studies where global wave-like rules were pre-specified to construct the actuation sequence of segments, this phase coordination method generates robot locomotion by assigning the actuation phase differences between adjacent segments without any global prerequisite rules. To effectively coordinate the phase differences, different symmetry properties are introduced. Optimization is then carried out on various symmetrically coordinated phase-difference patterns to maximize the average steady-state velocity of the robot. It is shown that the maximum average velocity is always achieved when the reflectional symmetry is included in the phase-difference pattern, and the identical-phase-difference (IPD) pattern is preferred for implementation because it reduces the number of independent phase variables to only one without significant loss in locomotion performance. Extensive analytical investigations on the IPD pattern reveal the relationship between the average locomotion velocity and some important parameters. Theoretical findings on the relationship between the average velocity and the phase difference in the IPD pattern are verified via experimental investigations on an 8-segment earthworm-like metameric robot prototype. Finally, this paper reveals an interesting result that the optimized phase-difference pattern can naturally generate peristalsis waves in metameric robot locomotion without global prerequisite wave-like rules.

Template for robust soft-body crawling with reflex-triggered gripping

Caterpillars show a remarkable ability to get around in complex environments (e.g. tree branches). Part of this is attributable to crochets which allow the animal to firmly attach to a wide range of substrates. This introduces an additional challenge to locomotion, however, as the caterpillar needs a way to coordinate the release of the crochets and the activation of muscles to adjust body posture. Typical control models have focused on global coordination through a central pattern generator (CPG). This paper develops an alternative to the CPG, which accomplishes the same task and is robust to a wide range of body properties and control parameter variation. A one-dimensional model is proposed which consists of lumped masses connected by a network of springs, dampers and muscles. Computer simulations of the controller/model system are performed to verify its robustness and to permit comparison between the generated gaits and those observed in real caterpillars (specifically Manduca sexta.).

Conformational Modeling of Continuum Structures in Robotics and Structural Biology: A Review

Hyper-redundant (or snakelike) manipulators have many more degrees of freedom than are required to position and orient an object in space. They have been employed in a variety of applications ranging from search-and-rescue to minimally invasive surgical procedures, and recently they even have been proposed as solutions to problems in maintaining civil infrastructure and the repair of satellites. The kinematic and dynamic properties of snakelike robots are captured naturally using a continuum backbone curve equipped with a naturally evolving set of reference frames, stiffness properties, and mass density. When the snakelike robot has a continuum architecture, the backbone curve corresponds with the physical device itself. Interestingly, these same modeling ideas can be used to describe conformational shapes of DNA molecules and filamentous protein structures in solution and in cells. This paper reviews several classes of snakelike robots: (1) hyper-redundant manipulators guided by backbone curves; (2) flexible steerable needles; and (3) concentric tube continuum robots. It is then shown how the same mathematical modeling methods used in these robotics contexts can be used to model molecules such as DNA. All of these problems are treated in the context of a common mathematical framework based on the differential geometry of curves, continuum mechanics, and variational calculus. Both coordinate-dependent Euler-Lagrange formulations and coordinate-free Euler-Poincaré approaches are reviewed.

Differential resistance feedback control of a self-sensing shape memory alloy actuated system

There is a growing trend towards miniaturization, and with it comes an increasing need for miniature sensors and actuators for control. Moreover situations occur wherein implementation of external physical sensor is impossible, here self-sensing lends its hand appropriately. Though self-sensing actuation (SSA) is extensively studied in piezoelectric, exploring this property in shape memory alloy is still under study. A simple scheme is developed which allows differential resistance measurement of antagonistic shape memory alloy actuated wires to concurrently sense and actuate in a closed loop system. The usefulness of the proposed scheme is experimentally verified by designing a one link manipulator arm and is performed in a real time tracking control. In a practical implementation of the self-sensing actuator a newly proposed signal processing electronic circuit is used for direct differential resistance feedback control upto a bandwidth of 1.8 Hz. The control design uses fuzzy PID which requires no detailed information about the constitutive model of SMA. At an operating frequency of 1 Hz, the result of the self-sensing feedback control with an angular tracking accuracy of ±0.06° over a movement range of ±15° is demonstrated.

Efficient worm-like locomotion: slip and control of soft-bodied peristaltic robots

In this work, we present a dynamic simulation of an earthworm-like robot moving in a pipe with radially symmetric Coulomb friction contact. Under these conditions, peristaltic locomotion is efficient if slip is minimized. We characterize ways to reduce slip-related losses in a constant-radius pipe. Using these principles, we can design controllers that can navigate pipes even with a narrowing in radius. We propose a stable heteroclinic channel controller that takes advantage of contact force feedback on each segment. In an example narrowing pipe, this controller loses 40% less energy to slip compared to the best-fit sine wave controller. The peristaltic locomotion with feedback also has greater speed and more consistent forward progress