From intelligent robot to multi-agent robotic system

The robot technology last decade had two streams of research, such as single but very intelligent robot and multiple agent type of robots. The robot research has been aimed at going to make a robot more flexible and more intelligent, using learning, adaptive and evolutionary capabilities. The work is mainly focused on the modelling and control enhanced by the computational intelligence, so that we can have robots how flexible and wide range of operation robot can engage in with less instructions. Sensor technology has also been improved to reduce the uncertainty, while actuator technology was assured to make the controller work as exactly as desired. On the other hand, the multiple autonomous robotic system has also been studied for cooperative/competitive works in the frame work of a group robotics, network robots, swarm intelligent robotic system or collective intelligent robotic system, believing that more robots, better work. One example is soccer robots. There are also lots of works in reality which might be carried out by multiple robots but single robots. But there are not many works which shows many intelligent robots work together in the real world. There are clearly many good works in both single and multiagent type of robots but apparently something missing in between them. Why is it difficult to make multiagent type of intelligent robots? we discuss this issue and show some future problems.