Most bio-inspired robots have been based on animals with jointed, stiff skeletons. There is now an increasing interest in mimicking the robust performance of animals in natural environments by incorporating compliant materials into the locomotory system. However, the mechanics of moving, highly conformable structures are particularly difficult to predict. This paper proposes a planar, extensible-link model for the soft-bodied tobacco hornworm caterpillar, Manduca sexta, to provide insight for biologists and engineers studying locomotion by highly deformable animals and caterpillar-like robots. Using inverse dynamics to process experimentally acquired point-tracking data, ground reaction forces and internal forces were determined for a crawling caterpillar. Computed ground reaction forces were compared to experimental data to validate the model. The results show that a system of linked extendable joints can faithfully describe the general form and magnitude of the contact forces produced by a crawling caterpillar. Furthermore, the model can be used to compute internal forces that cannot be measured experimentally. It is predicted that between different body segments in stance phase the body is mostly kept in tension and that compression only occurs during the swing phase when the prolegs release their grip. This finding supports a recently proposed mechanism for locomotion by soft animals in which the substrate transfers compressive forces from one part of the body to another (the environmental skeleton) thereby minimizing the need for hydrostatic stiffening. The model also provides a new means to characterize and test control strategies used in caterpillar crawling and soft robot locomotion.

Given the complexity of the problem, genetic algorithms are one of the more promising methods of discovering control schemes for soft robotics. Since physically embodied evolution is time consuming and expensive, an outstanding challenge lies in developing fast and suitably realistic simulations in which to evolve soft robot gaits. We describe two parallel methods of using NVidia's PhysX, a hardware-accelerated (GPGPU) physics engine, in order to evolve and optimize soft bodied gaits. The first method involves the evolution of open-loop gaits using a reduced-order lumped parameter model. The second method involves harnessing PhysX's soft-bodied material simulation capabilites. In each case we discuss the the challenges and possibilities involved in using the PhysX for evolutionary soft robotics.

SUMMARYWhen generating gaits for soft robots (those with no explicit joints), it is not evident that undulating control schemes are the most efficient. In considering alternative control schemes, however, the computational costs of evaluating continuum mechanic models of soft robots represent a significant bottleneck. We consider the use of lumped dynamic models for soft robotic systems. Such models have not been employed previously to design gaits for soft robotic systems, though they are widely used to simulate robots with compliant joints. A major question is whether these methods are accurate enough to be representations of soft robots to enable gait design and optimization. This paper addresses the potential “reality gap” between simulation and experiment for the particular case of a soft caterpillar-like robot. Experiments with a prototype soft crawler demonstrate that the lumped dynamic model can capture essential soft-robot mechanics well enough to enable gait optimization. Significantly, experiments verified that a prototype robot achieved high performance for control patterns optimized in simulation and dramatically reduced performance for gait parameters perturbed from their optimized values.

Abstract-In generating efficient gaits for biomimetic robots, control commands and robot morphology are closely coupled, particularly for soft bodied robots with complex internal dynamics. Achieving optimal robot energy consumption is only possible if robot control parameters and morphology are tuned simultaneously. Genetic Algorithms (GAs) are well suited for this purpose. In this application, however, GAs converge slowly because of the high dimensionality of the fitness landscape, the limited number of successful designs within this landscape, and the significant computational cost of evaluating the fitness function using dynamics simulations. To accelerate GA convergence for design applications involving biomimetic robots, a new physics-based preprocessing methodology is proposed. This preprocessing strategy was applied to develop gaits for a biomimetic caterpillar robot. Convergence speeds were observed to increase significantly through the application of the physicsbased preprocessing.

This paper introduces a metric called degree of shapeability to determine whether or not it is possible to assign a set of desired eigenstates to a particular linear dynamic system. Specifying an eigenstate, which consists of an eigenvalue and its associated eigenvector, can be essential in designing certain vibrating structures and undulating mechanisms. In general it may be desirable to specify more than one eigenstate. The number of eigenstates that can be specified for a particular linear system is not always immediately evident, however, because of physical and kinematic constraints, as well as actuator placement. To quantify the potential for active feedback and/or passive parameter tuning to influence a system’s eigenstates, we identify a metric, the degree of shapeability, that relates the structure of the linear system to the number of eigenstates that a designer may fully define. We investigate the effect of system equality and inequality constraints on the ability to specify eigenstates for linear dynamic systems.