Recently, the Evolutionary Robotics(ER) approach has been attracting lots of attention in the field of robotics and artificial life, since it can automatically synthesize controllers by taking the embodiment and the interaction dynamics with the environment into account. However, the ER approach still has several issues that have to be solved. One of the most serious issues in the ER approach is that it becomes significantly difficult to evolve the whole controller in one go(i.e. one-shot evolution), as the complexity of the desired task and/or the interaction dynamics with the environment increases. This is called the bootstrap problem. In order to alleviate this, it is highly indispensable to develop a method that can efficiently increase a controllers' ability without impairing the previously obtained ones(plasticity vs. stability dilemma). In this study, we propose a method for an incremental evolution of neurocontrollers by introducing a receptor-ligand concept, more specifically polymorphic property of neural circuits through diffusion-reaction mechanism of chemical substances, so-called neuromodulators. To investigate the effectiveness of our approach, we take a peg-pushing task, which requires an appropriate sequence of behavior to accomplish a task, as a practical example. We contrast this method with the conventional one in which simply the synaptic weights are the target to be evolved by carrying out simulations.
|Number of pages
|Published - 2001
|2001 IEEE/RSJ International Conference on Intelligent Robots and Systems - Maui, HI, United States
Duration: 2001 Oct 29 → 2001 Nov 3
|2001 IEEE/RSJ International Conference on Intelligent Robots and Systems
|01/10/29 → 01/11/3