In this paper, an approach for behavior imitation using visual information was introduced. The imitation process is done by a self organizing neural network module. From several demonstrations of task operation, a vision system captures movement of the demonstrator mobile robot and associated objects in an operation field. Then, the movement features are extracted to present to an imitation engine. Finally, skill or decision policy from teacher's demonstration is extracted and embedded into a self organizing neural network without explicit external supervisory signals. A simple action selection algorithm for choosing action from learned network is proposed. The algorithm was implemented and tested on a simulated robot and a real mobile robot to imitate two simple robot soccer behaviors: approaching the target and obstacle avoidance. Furthermore, the concept of similarity measure is introduced to evaluate imitation performance from the demonstrator

The effectiveness of multirobot learning in achieving optimal, cooperative solutions is potentially affected by various factors having to do with the nature and configuration of the robots and the nature and configuration of the robots and the nature of the learning entities. Varying one factor wrongly may lead to undesirable results. There is no reported work on how systematically to set up these factors. In this paper, we methodically test the effect of varying four common factors (reward scope, global information delay, diversity of robots, and number of robots) in a decentralized multirobot system, first in simulation and then on real robots. The results show that two of these factors, reward scope and global information delay, if set up incorrectly, can prevent optimal, cooperative solutions.

Learning can be an effective way for robot systems to deal with dynamic environments and changing task conditions. However, popular single-robot learning algorithms based on discounted rewards, such as Q learning, do not achieve cooperation (i.e., purposeful division of labor) when applied to task-level multirobot systems. A task-level system is defined as one performing a mission that is decomposed into subtasks shared among robots. We demonstrate the superiority of average-reward-based learning such as the Monte Carlo algorithm for task-level multirobot systems, and suggest an explanation for this superiority.