Manipulation tasks

From 3D scanning Knowledge base - Photoneo wiki
Jump to: navigation, search

The manipulation task can be phrased as "find-grab-place". To find the object of known shape we can use the Localisation SDK. This provides us the exact position and pose (angle, orientation) of the object. Now we need to convert these coordinates to a sequence of robot arm movement. First of all, we have to match robot coordinates to the camera coordinates, but this is easy since the calibration is done only once at the beginning. Second, we need to plan the arm movement so that it does not collide with other objects, walls, box or other items. We also need to define the gripper to make it possible for the robot to grab the object. The design of the gripper depends on the shape of the object, its weight, mechanical properties such as surface friction. Fragile objects or deformable objects require additional attention. There is a plenty options to choose from. After the object was grabbed, again the movement planning takes the place to transport the object at the desired destination.