A modular robot can be reconfigured and reorganized to perform different tasks. Due to the large number of configurations that this type of robot can have, several types of techniques have been developed to generate locomotion tasks in an adaptive manner. One of these techniques transfers sets of parameters to the robot controller from a simulation. However, in most cases the simulated approach is not appropriate, since it does not take into account all physical interactions between the robot and the environment. This paper shows the design of a flexible controller that adapts to the different configurations of a modular chain-type robot, which coordinates the movements of the robot using a Central Pattern Generator (CPG). The CPG is integrated with an optimization algorithm to estimate sets of movements, which allow the robot to navigate in its environment autonomously from the information of sensors and in real time.
|Konference||Ibero-American Conference on Artificial Intelligence|
|Periode||13/11/2018 → 16/11/2018|
|Navn||Lecture Notes in Computer Science|