Robots that have been optimized in simulation often underperform the real world comparison to their simulated counterparts. This difference performance is called a reality-gap. In this paper, we use two methods, genetic algorithm and topology optimization, optimize quadruped robot. We look at original robots’ reality compare results. Both methods show improvement robot’s efficiency, however opt...

We are developing a general quadrupedal locomotion controller by using a neural model involving a CPG (Central Pattern Generator) utilizing normal ground reaction force (leg loading) sensory feedback for both rolling motion (posture) and gait (rhythmic motion) control. In this abstract, we report the results of experiments using a quadruped robot “Kotetsu” in order to verify the results of our ...

We have proposed a self-reconfigurable robotic module, which has a very simple structure. The system is capable of not only building static structure but generating dynamic robotic motion. We also developed a simulator for motion planning. In this paper, we present details of the mechanical / electrical design of the developed module and its control system architecture. Experiments using ten mo...

The purpose of this paper is to describe two cases of a hitherto unknown syndrome superficially resembling myotonia. Detailed studies have helped to separate this syndrome of continuous musclefibre activity from a group of 100 myotonic subjects, and it will be referred to in this paper as the 'syndrome of continuous muscle-fibre activity'. This descriptive label has been chosen, though the more...

In this paper, motion patterns of a giant entertainment quadruped robot’s head, body and tail parts are analyzed and extracted. Then, a fuzzy controller is designed in order to obtain life-like motion. Motion data are derived using 3D animation during robot’s walking and an organization mechanism is produced to obtain this kind of data. This paper introduces a data extraction method using a 3D ...

In this paper we suggest a novel method to approximate the fitness function of a genetic programming approach in order to develop fast and stable gait patterns for a quadruped robot. Therefore, gait patterns are classified by so called Signal Space Detectors. We show how a Signal Space Detector can extract information about the reliability of a classification. Finally, we demonstrate how this i...