Spine dynamics as a computational resource in spine-driven quadruped locomotion

Abstract

Recent results suggest that compliance and nonlinearity in physical bodies of soft robots may not be disadvantageous properties with respect to control, but rather of advantage. In the context of morphological computation one could see such complex structures as potential computational resources. In this study, we implement and exploit this view point in a spine-driven quadruped robot called Kitty by using its flexible spine as a computational resource. The spine is an actuated multi-joint structure consisting of a sequence of soft silicone blocks. Its complex dynamics are captured by a set of force sensors and used to construct a closed-loop to drive the motor commands. We use simple static, linear readout weights to combine the sensor values to generate multiple gait patterns (bounding, trotting, turning behavior). In addition, we demonstrate the robustness of the setup by applying strong external perturbations in form of additional loads. The system is able to fully recover to its nominal gait patterns (which are encoded in the linear readout weights) after the perturbation has vanished.1 Reservoir computing is a machine learning technique used to emulate complex, nonlinear computations by employing a randomly initiated (but afterwards fixed in their parameters) complex, nonlinear dynamical network of nonlinear dynamical systems (i.e., the reservoir). For more details we refer to [8].

Pneumatic Reservoir Computing for Sensing Soft Body: Computational Ability of Air in Tube and Its Application to Posture Estimation of Soft Exoskeleton

“…The separation property means that the input signals that should be distinguished are mapped to different physical states. The dynamics of soft materials, (3)(4)(5)(6) fluids, (7,8) mass-spring systems, (14,15) electronic circuits, (16) optical circuits, (17) memristors, (18) spintronic systems, (19) quantum systems, (20) and so forth have been exploited as computational resources for physical reservoir computing, and it has been experimentally shown that the systems have computational ability in benchmark tasks. (12,13) (b) physical reservoir computing using a mass-spring system, (14) and (c) proposed physical reservoir computing using a pneumatic pipeline system.…”

“…In the control of some robots, locomotion was controlled by using signals from sensors embedded in the soft part of the body. (3,4) It has been shown that a soft arm resembling an octopus arm can be exploited for some computational tasks, such as a parity check. (5,6) These methods, however, use some movements of the robot itself for computation and require a specially designed body morphology [e.g., tensegrity structure (4) ] to be used in combination with the robot's primarily required motion (e.g., locomotion).…”

Intelligent design for quadruped robot on a dynamic flexible surface with an active spine

“…5 The developed model has possible applications in fields such as search and rescue operations, military sector, mining, construction, and robotics for autonomous exploration and monitoring. 6 Additionally, the results could help researchers understand better the dynamic behavior of legged robots, contributing to future designs' improvement.…”

Morphologically induced stability on an underwater legged robot with a deformable body

“…The idea has a wide number of applications within its scope. MC is sometimes referred to as the possibility of using the complex nonlinear dynamics of a deformable body to serve as a computational resource in control tasks Nakajima et al (2014Nakajima et al ( , 2015; Hauser et al (2018), with a particular focus on soft manipulators control Nakajima et al (2013); Eder et al (2017) and locomotion Zhao et al (2013). A different branch of MC can be found in the field of robot locomotion, where the interaction with a complex and nonstructured environment is not handled by means of complex controllers, but by harnessing the intrinsic properties of the materials used in construction Auerbach and Bongard (2017); Calisti et al (2017).…”