Dwarf intelligence

We have been studying multiple mobile robot system from 1989. This system can realize high quality tasks by cooperation of robots. However, it is difficult to construct a multi-robot architecture that can deal with general tasks. We consider that a practical multiple mobile robot system can be established by combining rough, robust off-line plan-making for robots, and on-line navigation of robots using sensor information. We call this “dwarf intelligence.”

It is natural to consider that “motion” is essential for mobile robots. Here we classify robot tasks from the viewpoint of “how are the dimensions of task demands?” and “how many times should tasks be made?” Tasks can be classified into four by combination of the above two issues. Classification and examples of tasks are shown in Table 1. The final target of this study is to construct an architecture for multiple mobile robot system that can realize not only these four tasks but also compounded tasks consisting of the four tasks.

We are doing studies on “robust sweep task planning and real time navigation,” “iterative transportation planning” and “autonomous tasks assignment among robot in doing cooperative transportation.” The first two topics correspond to two classes of the ones at Table 1. The last topic deals with task assignment among multiple mobile robots when there are many sub-tasks to be realized.


  1. Tamio Arai, Jun Ota: Dwarf Intelligence -- A Large Object Carried by Seven Dwarves --, Robotics and Autonomous Systems, 18, 149/155(1996).
  2. Tamio Arai, Jun Ota: Let Us Work Together --- Task Planning of Multiple Mobile Robots ---, In Proc. 1996 IEEE/RSJ Int. Conf. Intell. Robots and Systems (IROS'96), 1,298/303(1996).


Table 1 Classification of Robot Tasks






Iterative transportation




Collecting samples