Inducing Known Movements and Emerging Adaptive Movements

(Tsukahara Y., Kuniyoshi Y., Ann. Conf. of Robotics Soc. of Japan, 2008)

The emergence of a robot’s movement in an unknown environment is a key to avoid the robot from failure in its behavior. An emergence model based on coupled chaotic fields (CCF) was proposed, which can instantly transit its behavior to new one in response to sudden change of the environment. We propose an extended emergence model to induce known movements. First, we studied the characteristics of the emergent movements. Next, we studied the constitution method for CCF, which is used as bias to induce the behavior. Finally, we experimentally show the possibility of imposing a known movement pattern to the system and then releasing it to resume the emergent adaptivity at desired timings.

<The structure of the insect-like robot and experimental environment>

<Coupled chaotic system>

Coupled Map Fields is implemented in the following form.

Where, denotes the intemal state of th element at time the total number of elements, and  the connection weight between elements.  can be any chaos function.  Here, we adopt a standard logistic map represented as the following.

<The emergence of movements (500 trials) and representative direction>

<The transition of dynamics pattern in response to collision>

<The controllability of movements>

[Model for behavior emergence/adaptation as coupled chaotic field]

[The connection weight as a bias induces known movement]

[The adaptivity before/after the bias release]

[The control of movements with Bias switching]

Share