3D Vision-Guided casting workpiece loading
Project background: Die casting is a typical flat workpiece with few effective identification features and large rigidity difference. Traditional visual positioning is prone to identification deviation, identification error, and inaccurate ranging. However, such application conditions require extremely high process accuracy and put forward extremely high requirements for the accuracy, category compatibility and continuous operation stability of visual positioning guidance.
- Project advantages:
- Training deep learning models based on DexForce self-developed DexVerse™ embodies intelligent engine, which can identify various sizes and shapes of workpieces, without repeated model training, and has strong applicability;
- Support to configure multiple product models at the same time, and switch operations according to robot signals; There is no need to stop production to collect annotated data, and the deep learning model training of the new workpiece can be completed in 8 hours, and the production can be quickly launched.
- Zero code, zero programming, direct adjustment of parameters can achieve visual positioning, the fastest 5 minutes to complete visual configuration, 20 minutes to cooperate with the robot to grasp debugging.
- Application results
Based on the DexForce intelligent 3D vision system, it can automatically complete the identification, positioning and processing of metal parts, reducing the frequency and error rate of manual operation. Through algorithm optimization and intelligent control, it can quickly and accurately identify metal parts, realize automatic loading and unloading operation, reduce labor costs and production line maintenance costs, so as to achieve the purpose of cost reduction.
