Once you have identified which elements of a task will be controlled by MIRAI, you will need to decide which type of MIRAI skill is best suited for your program. MIRAI has two classes of skills:
Positioning skills allow you to position the robot's tool center point (TCP) in real-time relative to a visible target (object for picking, work piece for insertion, etc.). The robot takes the most direct/efficient path to the target position.
To properly configure a positioning skill you will need to consider the following:
- The path between the starting point of the MIRAI skill execution (the handover point where MIRAI takes control) to the target position must be free of obstructions.
Examples using Positioning Skills:
- Object picking (with variance in color, size, shape, position)
- Positioning a workpiece precisely for an insertion task
- Placing a sensor/tool relative to a work piece for quality inspection
Typically, positioning skills require less time and training compared to motion skills and the AI can easily achieve high comprehension with fewer recorded episodes. We recommend at least 20 episodes for most positioning skills. (See section 2.4 ‘Training Positioning Skills’ for more details).
Motion skills allow the robot to perform more complex movements in real-time. This is useful for a very wide range of applications, such as insertions, following a contour, picking objects from a moving conveyor, or tool positioning in crowded situations where a direct path from the start point to the target point is not normally feasible.
Examples using Motion Skills:
- Cable plugging (ribbon cables, JST connectors, etc)
- Following a curve or contour, such as tracing the surface of an airplane wing
- Picking or placing objects to and from a moving conveyor belt
- Finishing tasks such as polishing or gluing
- Inserting a stick of RAM into a motherboard
Motion skills are more demanding in terms of training requirements. We recommend at least 50-100 episodes depending on degrees of freedom being used. It is essential to training this type of skill in as many variations and repetitions as possible to achieve a desired AI comprehension.
For more details, check out the video below: