Theoretical Modeling of a Pressure-operated Soft Snake Robot

This paper addresses the theoretical modeling of the dynamics of a pressure-operated soft snake robot. An accurate dynamic model is a fundamental requirement for optimization, control, navigation, and learning algorithms for a mobile robot that can undergo serpentine locomotion. Such algorithms can be readily implemented for traditional rigid robots, but remain a challenge for nonlinear and low-bandwidth soft robotic systems. A framework to solve the 2-D modeling problem of a soft robotic snake is detailed with a general approach applicable to most pressureoperated soft robots that are developed by a modular kinematic arrangement of bending-type fluidic elastomer actuators. The model is simulated using measured physical parameters of the robot and workspace. The theoretical results are verified through a proof-of-concept comparison to locomotion experiments on a flat surface with measured frictional properties. Experimental results indicate that the proposed model describes the motion of the robot. INTRODUCTION Robots promise to improve our lives in search and rescue applications. These applications require special robotic capabilities that may not be fulfilled by traditional mobile robots such as operating through narrow openings or complex passages. For ∗Address all correspondence to this author. such conditions, a robotic snake is a suitable candidate since it can navigate on unstructured terrain without limbs while being able to pass through narrow space similar to its biological counterpart. Many researchers analyzed the principles of snake locomotion and developed robotic equivalents that can replicate snake motion. The first snake robot was developed by Shigeo Hirose at Tokyo Institute of Technology in 1971 [1]. During 40 years of research since, many snake robots have been developed, including Anna Konda, a large firefighting snake, Aiko, a portable system for experimentation, and Pneumosnake, developed to investigate joint actuation based on pneumatic bellows [2]. Recent research on the snake inspired robots for the minimally invasive surgery application [3], the snake robot can work on a step environment [4]. On the other hand, current snake robots do not utilize body flexibility. Since traditional robot fabrication is based on rigid links, robotic snakes may not be as safe and adaptive as their natural counterparts. Our objective in this research is to develop a pneumatically actuated soft robotic snake that can overcome the limitations of rigid snake robots. Soft robotics has recently seen a flurry of research including many different kinds of crawling robots [5–7]. The first generation of our soft snake robot was developed in [8] and [9]. The body is fabricated by molding in three layers. The total manufacturing and assembly process takes 14 hours 1 Copyright c © 2014 by ASME FIGURE 1. Experimental prototype of our pressure-operated soft robotic snake. from scratch, resulting in an inexpensive robot. A recent prototype of our fluidic soft robot is shown in Fig. 1. Some challenges with the first iteration of the snake robot included the need for an accurate model for deeper research, a perception system for gait control, and a skin that offers anisotropic friction to eliminate the passive wheels, a current problem in snake robots in general. This paper focuses on the first challenge. Snake robot modeling is a mature discipline for rigid robots. [10] and [11] study modeling a rigid snake robot in 2-D. [12] add expressions for the linear velocity of individual links based on previous work and divide the general model into an actuated and an un-actuated part. Subsequently, partial feedback linearization of the model is presented. In addition, [12] proposes a simplified model after linearization and gives proofs of stability and controllability of a rigid snake robot based on the proposed model. On the other hand, [13–15] study segmented rigid snake robot modeling in 3D by taking vertical motions into account. However, there exists limited mathematical modeling studies for soft robots since the deformable nature of such systems creates a challenge, such that a soft body may create infinite degrees of freedom. In previous work, we utilized a fundamental constant curvature kinematic model and augmented an anisotropic friction function to iteratively describe the shape of the body over time and provide intuition about the locomotion of our soft snake robot [9]. In this work, we treat each soft segment as a joint and analyze short rigid connectors as links. This approach is compatible with existing rigid snake robot kinematics modeling studies and provides a more accurate description of the whole system. The outline of the paper is as follows: Section II shows the mathematical details of the soft snake robot model, Section III displays the dynamic simulation studies of a simplified model. Section IV describes the fabrication of the robot, the experimental setup and results. Section V concludes the paper and discusses potential future research directions. TABLE 1. Parameters of the Soft Robotic Snake Model