Quadrupedal locomotion: GasNets, CTRNNs and Hybrid CTRNN/PNNs compared
OPAL (Open@LaTrobe) (La Trobe University)Published 1 September 2004Open access
Phil Husbands, Gary McHale
Citations15
Generate an AI Snapshot to get a quick, structured summary of this paper.
Study Snapshot
ObjectiveStudy objective
MethodsResearch methodology
PopulationPopulation studied
Sample sizeSample sizes
OutcomesStudy outcomes here
ResultsStudy results comes here
LimitationsResearch study limitations comes here
A concise AI-generated summary of the paper will appear here once you click Generate AI Snapshot.
Abstract
Evolutionary Robotics seeks to use evolutionary techniques to create both physical and physically simulated robots capable of exhibiting characteristics commonly associated with living organisms. Typically, biologically inspired artificial neural networks are evolved to act as sensorimotor control systems. These networks include; GasNets, Continuous Time Recurrent Neural Networks (CTRNNs) and Plastic Neural Networks (PNNs). This paper seeks to compare the performance of such networks in solving the problem of locomotion in a physically simulated quadruped. The results in this paper, taken together with those of other studies (summarized in this paper) help us to assess the relative strengths and weaknesses of the these three different approaches.
Keywords
Computer ScienceEngineering
Adaptive BehaviorOn the Dynamics of Small Continuous-Time Recurrent Neural Networks
398 Citations1995Randall D. Beer
This article begins a systematic examination of the dynamics of continuous-time recurrent neural networks with a fairly complete description of the possible dynamical behavior and bifurcations of one- and two-neuron circuits.
IEEE Transactions on Evolutionary ComputationEvolution of central pattern generators for bipedal walking in a real-time physics environment
182 Citations2002Torsten Reil, Phil Husbands
Using a full rigid-body simulation of a biped, it was possible to evolve recurrent neural networks that controlled stable straight-line walking on a planar surface and simple sensory input to locate a sound source was integrated to achieve directional walking.
Journal of Mathematical BiologyModels of central pattern generators for quadruped locomotion
160 Citations2001Pietro-Luciano Buono, Martin Golubitsky
It is shown that under mild assumptions on the cells and the coupling of the network, primary gaits can be produced from Hopf bifurcation by varying only coupling strengths of the Network of symmetrically coupled cells.
Repository for Publications and Research Data (ETH Zurich)Evolution of Plastic Neurocontrollers for Situated Agents
118 Citations1996Dario Floreano, Francesco Mondada
A novel approach to the evolutionary development of autonomous situated agents based on the assumption that the neural mechanisms underlying ontogenetic learning are themselves developed and shaped by evolutionary process is investigated.
Biologically inspired neural controllers for motor control in a quadruped robot
73 Citations2000Aude Billard, Auke Jan Ijspeert
Sets of artificial neural networks are presented which provide pattern generation and gait control, allowing continuous passage from walking to trotting to galloping, and control of sitting and lying down behaviors, and 3) control of scratching.
Adaptive BehaviorNeuromodulated Control of Bipedal Locomotion Using a Polymorphic CPG Circuit
50 Citations2003Akio Ishiguro, Akinobu Fujii +1 more
The results show that the biped robot successfully copes with environmental perturbation by dynamically changing the torque outputs applied to the joints, and the proposed approach outperforms a monolithic CPG model with sensory feedback.
Neural ComputationCenter-Crossing Recurrent Neural Networks for the Evolution of Rhythmic Behavior
48 Citations2002Boonyanit Mathayomchan, Randall D. Beer
It is demonstrated that relative to a random initial population, seeding the initial population of an evolutionary search with center-crossing networks significantly improves both the frequency and the speed with which high-fitness oscillatory circuits evolve on a simple walking task.
The MIT Press eBooksLevels of Dynamics and Adaptive Behavior in Evolutionary Neural Controllers
46 Citations2002Jesper Blynel, Dario Floreano
Two classes of dynamical recurrent neural networks are compared on two behavioral tasks aimed at exploring their capabilities to display reinforcement-learning like behaviors and adaptation to unpredictable environmental changes, but PNNs display significantly better performance when sensory-motor re-adaptation is required after the evolutionary process.
Lecture notes in computer scienceMaking Evolution an Offer It Can’t Refuse: Morphology and the Extradimensional Bypass
37 Citations2001Josh Bongard, Chandana Paul
It is shown that stable walking is achieved through coupled optimization of both the controller and the mass ratios and mass distributions of the biped, and the inclusion of the correct set of morphological parameters improves the evolution of adaptive behaviour in simulated agents.
OPAL (Open@LaTrobe) (La Trobe University)GasNets and other evovalble neural networks applied to bipedal locomotion
30 Citations2004Phil Husbands, Gary McHale
Of the 14 distinct networks tested, GasNets were the only network to achieve cyclical locomotion, although CTRNNs were able to attain a higher level of average fitness.
Perspectives in neural computingBrains, Gases and Robots
14 Citations1998Phil Husbands, Tom Smith +4 more
Although the picture of biological neuronal networks changes every few years, contemporary neuroscience can provide a rich source of inspiration in devising alternative styles of artificial network, which are generally very different kinds of systems from those that generate sophisticated adaptive behaviours in animals.
Lecture notes in computer scienceEvolutionary Robots with Fast Adaptive Behavior in New Environments
14 Citations2000Joseba Urzelai, Dario Floreano
The results show that evolutionary adaptive controllers can adapt to environmental changes that involve new sensory characteristics (including transfers from simulation to reality) and new spatial relationships.
Theory in BiosciencesVolume signalling in real and robot nervous systems
12 Citations2001Phil Husbands, Andy Philippides +2 more
It is shown that the morphology of the neuromodulator source plays a highly significant role in the diffusion patterns observed, and the virtual diffusing modulators are found to provide significant advantages.