To see movies, please visit our new site.

The second prototype M-TRAN II has enough power for whole body motions, such as locomotion. As the modular robot changes its configuration, designing a locomotion pattern should be automated. We have developed a program for this pattern generation, which uses CPG (Central Pattern Generator) network as a dynamic pattern generator and uses GA (Genetic Algorithm) for optimization. We modeled M-TRAN dynamics in the host computer and made repetitive dynamics simulation according to the GA process, and optimized locomotion patterns for several configurations of M-TRAN. Generated patterns are verified by hardware experiments. (->details)
In the two experiments (10) and (11) of the following, locomotions and self-reconfiguration are combined.

	Simulation		Experiment
(1) 4-Legged Walking Before evolution by GA
(2) 4-Legged Walking After evolution by GA
(3) Turning round
(4) Mini- 4-legged walker
(5) Double-thread Caterpillar motion (H-shape)
(6) Six-legged walker
(7) Single-thread Caterpillar motion
(8) Wheel (Crawler?)			unsuccessful?
(9)Spider?
(10) Reconfiguration from H-shape to Single-thread
(11) 4-Leg walk -> H-shape -> Single-thread Caterpillar
(12) H-Shape -> Crawler -> 4-Leg walk -> H-Shape -> Single-thread Caterpillar
(13) Axisymmetric 4-Leg -> Straight line
(14)Sidewinder
(15) Walking around
(16) Climb
(17) 4leg ->Caterpillar (2003/11/13)
(18)Union (2004/Jan)