The last week, I have started a new assignment, determining the dexterous position of the KUKA youBot. This exploration is done through a numeric and systematic exploration of the joint values. This evening I have the first picture of the result.
I have done a first program to compute the position of the end-effector for each joint values. All values are stored in a dict and then dumped into a flat file for further exploration. For a step of 0.05 rad, the program generates 310.000 values in a couple of seconds. A subsequent program loads the file and plot the number of solutions for a given position. The warmer, the higher the number of solutions.
KUKA youBot workspace exploration (number of solutions by end-effector position)
The big dark shape on the bottom left if the base of the robot.
The first conclusion, which is not really a surprise, is that the area just in front of the youBot and close to the ground is very dexterous. For the manipulation of an object or a toy on the ground, the youBot should try to position itself close to it, between 100 and 150 mm. The area around the plate, at -200 on the x axes and just above the base, it not very dexterous. So the manipulation of something on the plate will not be that easy. The arm may have to rotate around the shoulder axes for a better ease.
Something very useful and not obvious is the area at 400 mm above the ground. That means that the robot should be able to manipulate objects at that height, which is approximately the height of a chair. Here again, the arm should probably be rotated around the shoulder to manipulate something at that position and in the front of the robot.
It is only a partial conclusion, as the plot is the number of solutions, not the number of directions, which the true indicator of the dexterity of the gripper.
All programs are available on this repo on GitHub.
The youBot in exploration of a very dexterous area
The post First result of the youBot workspace exploration appeared first on Zombie Brainz' Juice.
