Robotic arm driven by image analysis

Unlike predefined robotics programs, with the use of machine learning and computer vision methods, the robot is able to reach a die anywhere in its workspace. This technology brings flexibility and autonomy to the robotic arm and could for example be useful in automated sorting of parts, fruit or waste on a conveyor.

​

3 main steps allow the acquisition of the die:

​

1. The camera detects the die and calculates its centroid with respect to its own reference frame.

​

2. The coordinates are transformed according to the robot's reference frame. They are then transmitted and the robot is activated.

​

3. The arm is moved to the transmitted coordinates and the die is recovered.

 

In order to show the ability of the model to acquire the die at random positions, the die is successively caught and released by the robot without human intervention.