In order to reduce difficulties in carrying . out control experiments of an actual robot in a real environment and to speed up the development of the autonomous control algorithms for a mobilerobot, this paper proposes a mobile robot control system using a virtual environment, which consists of an actual robot hardware and a model environment assumed around the robot.
By measuring the robot motion in real time, the system knows the position of the robot in the model environment, simulates the data the robot would obtain with its external sensors if the environment were real and sends it to the robot. The robot is autonomously controlled in the model environment using the external sensor data. Because the environment is a model, there is no trouble such as collisions with real objects and damages of the robot hardware, even if the control of the robot fails. It is also easy to evaluate the control algorithms using the real time motion measurement of the robot.
We constructed an experiment system using the virtual environment and controlled the robot in it. We developed a collision avoidance algorithm against a wall and a method in which the position estimation error of dead reckoning is corrected based on wall recognition using the range information of ultrasonic sensors. Thus, we demonstrated the effectiveness of the system.
抄録全体を表示