Goal-directed multimodal locomotion through coupling between mechanical and attractor selection dynamics

Abstract

One of the most significant challenges in bio-inspired robotics is how to realize and take advantage of multimodal locomotion, which may help robots perform a variety of tasks adaptively in different environments. In order to address the challenge properly, it is important to notice that locomotion dynamics are the result of interactions between a particular internal control structure, the mechanical dynamics and the environment. From this perspective, this paper presents an approach to enable a robot to take advantage of its multiple locomotion modes by coupling the mechanical dynamics of the robot with an internal control structure known as an attractor selection model. The robot used is a curved-beam hopping robot; this robot, despite its simple actuation method, possesses rich and complex mechanical dynamics that are dependent on its interactions with the environment. Through dynamical coupling, we will show how this robot performs goal-directed locomotion by gracefully shifting between different locomotion modes regulated by sensory input, the robot's mechanical dynamics and an internally generated perturbation. The efficacy of the approach is validated and discussed based on the simulation and on real-world experiments.

An Information-Theoretic Approach to Self-Organisation: Emergence of Complex Interdependencies in Coupled Dynamical Systems

“…Let us consider a system composed by N parts or subsystems, which we call “agents” adopting the terminology from the robotics and multi-agent systems literature. However, these agents could correspond to different coordinates of the spatial movement of a single entity [ 45 ], or to sub-systems of heterogenous nature. The set of possible states for the k -th agent is denoted as , and hence the set of possible configurations of the system is , henceforth called “phase space.” The configuration of the system at time is determined by the vector , where is the corresponding state of the k -th agent and T is a collection of time indices.…”

A self organization approach to goal-directed multimodal locomotion based on Attractor Selection Mechanism

“…The question here is how to design a particular control architecture such that the self organization of the body-environment dynamics will result in the emergence of suitable locomotion modes depending on the task, without any previous knowledge about the robot's body or the environment. In our previous research, the ASM dynamics were designed with preceding knowledge about the robot's dynamics [17][18]. Here, it will be shown that an automatic exploration of different body-environment dynamics and the ability to stabilize onto particular attractors, which corresponds to distinguishable locomotion modes relevant to the task, can be enabled through a suitable control structure based on ASM.…”

“…the contact points can be placed slightly below the ground, and there is no traction force, except the gravitational force, when the contact points are above the ground. The model has been shown to adequately describe the interaction with the environment in the real robotic system [18]. It is assumed that ground reaction force is exerted on the 4 points of the foot: G p (p = 1, 2, 3, 4).…”

“…Despite the simplicity, it will be shown that the approach enables a robot to automatically explore different body-environment dynamics and stabilize onto particular attractors which correspond to locomotion modes relevant to the robot's goal. The robot used in this paper is a curvedbeam hopping robot, which despite its simple actuation method, possesses rich and complex dynamics depending on its mechanical configuration and environment [17][18][19].…”

Morphological Computation: Synergy of Body and Brain

“…Examples are the spine-driven robot [7], which uses the spine dynamics as part of its controller and the dynamics of an octopus arm that can be used for computation [25]. Within this first approach, there are also several works that discuss the importance of a tight body-brain-environment coupling, of which the following are just a few examples [13,[26][27][28][29][30][31]. Although very intuitive and compelling, this approach does not allow to quantify how the body reduces the computational burden of the brain.…”