In the wheeled mobile robot system with single scanning laser range sensor, since eyesight of the robot is vertically thin, inferior obstacles can not be founded by the robot. This problem can be improved by adding another range sensors or swing the sensor. But this approach is expensive and complex.
"IF THEY HAVE NO 3D VIEW, LET THEM LOOK DOWN!"
In this demonstration, robot finds inferior obstacles and take it to his nest box. Pitch angle of the robot is calculated from gyro sensor signal(ENC-03RC by muRata, 200yen at Akiduki Denshi Akihabara). And point data of scanning laser range sensor(URG-04LX by HOKUYO AUTOMATIC) are coordinate-transformed by the robot position and posture. Transformed data are recorded in the grid-map. Cells of grid-map, which have points with Z-position of 5cm to 10cm and don't have point over Z-position of 10m, are segmented and each segment are recognized as obstacle.
While the robot pushes obstacles to the box, its odometry position are modified according to the measured position of the nest box.
In this demonstration, the robot itself Googles a keyword that was given by the user in the Internet, and actually search for the real object that matches the keyword well in the real world. The robot does not have the information about the keyword object previously, but will search it by the Google Image Search. Comparing the color information of the image found in Google Image Search with the one captured in the real world using the USB camera mounted on the robot, the robot will judges whether the object in front of its camera is the thing the user is looking for.
These Autonomous Mobile Robots move around in the room. Usually, the robots avoid obstacles. When one of the robots finds the marker, the robot calls other two robots. The main points are the sound effects and multi robot cooperation using WLAN.
You can have a robot to draw any road sign of your favorite idea. How wonderful harmony is! First of all, you design your traversable trajectory using a computer. And next, your traversable trajectory should be divided into circular arc and straight line and set them to path to draw on a road for the robot. Next, the robot moves to starting position of drawing area on determined coordinate system by detecting position of two poles using URG sensor as a reference. The robot drops down a drawing pen at starting position, and tracks the planned path to draw a road sign. A lot of my time was spent to obtain smooth curve from the traversable trajectory by optimize combination of circular arc and straight line.
This training aims to train the ability in shifting the responsibility. There are two player in each end of the field. When robot comes to your area, please hit the robot by a board. when the robot reflects like hockey at a wall and a board, the count decreases. and when count reaches 0, the robot will explosion. The person who hit the robot at last has the responsible for this explosion. So you should not hit the robot at last. The ability in shifting the responsibility can be trained by this process.