|Budget Amount *help
¥2,600,000 (Direct Cost : ¥2,600,000)
Fiscal Year 1997 : ¥800,000 (Direct Cost : ¥800,000)
Fiscal Year 1996 : ¥900,000 (Direct Cost : ¥900,000)
Fiscal Year 1995 : ¥900,000 (Direct Cost : ¥900,000)
In this research, a method of behavior control based on the consciousness is proposed. In the conventional control scheme for robots, the execution and high-level information processing such as an environment recognition and a motion planning are clearly separated. That is to say, the environment recognition and motion planning are done in thinking time and the motion is simply controled by tracing the planned trajectories in the execution time. This means that the consciousness is lost in the execution time.
In this research, we tried to make the environment recognition and motion planning very fast, and to put them into the control loop of the task. As a result, we succeeded in realizing a behavior control system that is always conscious of the situation of the environment. Specifically, a control system that repeats a control action of the recognition, planning and motion execution at the rate of 10 times a second is developed.
For experiments a small-sized, omni-directional mobile robot has developed. In the experiment the mobile robot carried out the task of moving to a goal without collision in the dynamic environment. It is verified by experiments that mobility of the robot controled by the proposed method is done without failure except that a human handler intentionally disturbed the movement of the robto.
As a real-time vision to be used for the environment recognition, a mark-based vision is proposed to be implemented. By attaching a combination of marks and a barcode to each object, the vision enables the recognition of arbitrary number of 2.1/2 dimensionally-shaped objects in the 3-D environment.
In conclusion, we succeeded in realizing a kind of "robot consciousness" by making an environment recognition and motion planning real-time. It is experimentally shown that the control system based on the consciousness enables a very robust task execution.