ummary： Automatic loading and unloading of rubber hoses through AI+3D technology helps end customers improve production efficiency and quality and create flexible production solutions.

 Case background

The hose is placed in the material frame in a disordered manner, 3D vision identifies and locates the workpiece, guides the robot to grasp and load the hose in a disordered manner, replaces the manual with a mechanical arm, and the grasping accuracy is ±0.25mm, which can run stably for 24 hours.

Technical difficulties

1. The workpiece is pure black, which has a great impact on the camera point cloud imaging;

2. High accuracy requirements, accuracy requirements ±0.25mm;

3. The incoming materials are in disorder and need to be positioned twice;

4. A single robot is responsible for loading and unloading the workpiece of two stations.

 Solution

01. Using multiple exposure point cloud superposition to ensure the quality of the workpiece point cloud;

02. Adopt self-developed high-precision 3D visual positioning algorithm, workpiece positioning accuracy ±0.25mm;

03. Adopt deep learning and hardware acceleration algorithm to improve the tempo;

04. It can adapt to the scene of simultaneous production of multiple workpieces.