Multi-Station, Dual Robot, Four Pallet Positions: How One 3D Vision System Handles Complex Feeding

One 3D camera, corresponding to two robots and four pallet positions, simultaneously identifying three different materials—evaporators, black foam, and trays—this is not an 'extreme challenge' but a real deployment by Migrate Technology at an automotive parts manufacturing site.

01 Site Conditions

The project involves multi-station linkage: 1 camera → 2 robots → 4 pallet positions, with diverse materials, strict placement rules, and demanding precision requirements. Handling objects: 50 types of evaporators, 1 type of black foam, 7 types of trays Stacking method: Alternating layers of evaporators and foam, with the top layer always being an evaporator Pickup requirements: Collision avoidance, layer-by-layer identification, model verification, and abnormal alarm Challenges: Black foam is easily deformed, evaporators have dimensional tolerances of ±3mm, and tray positioning requires high precision; traditional teaching and 2D vision cannot reliably handle these.

02 Workflow

Migrate Technology deployed the Epic Eye Laser L 3D camera, mounted on a moving module that can switch between four pallet positions (two tray pallets and two evaporator/foam pallets) upon PLC commands.

Specific operation flow: Tray identification: The camera moves to the tray pallet, identifies the pose, and guides Robot 1 to pick and place it on the conveyor line. Evaporator identification: The camera switches to the evaporator pallet. The AI algorithm first determines whether the top layer is an evaporator or foam. If it is an evaporator, it precisely calculates the pose and guides Robot 1 to pick and accurately place it into the tray. Foam identification: After the evaporator is removed, the camera takes a second image to identify the quantity and pose of the black foam on the current layer, guiding Robot 2 to sequentially pick and stack. Cycle operation: This process repeats until the pallet is empty.

This entire process realizes one vision system driving two robots to handle three types of materials, significantly reducing hardware costs and deployment complexity. Thanks to the excellent ambient light interference resistance and superior imaging performance of the Epic Eye Laser L for dark objects, even the light-absorbing black foam yields complete and clear point cloud data, laying a solid foundation for precise AI algorithm calculations.

03 Value Delivered

High Precision ±0.3mm recognition accuracy and ±0.5mm pickup accuracy ensure that evaporators, trays, and foam are picked and placed correctly at once, effectively avoiding production line issues such as jamming, misalignment, and collisions, greatly reducing material loss and equipment downtime.

High Efficiency The entire vision system's single recognition cycle is controlled within 3.5 seconds. The efficient workflow of the Epic Eye Laser L camera significantly reduces waiting time and greatly improves operational efficiency.

High Flexibility One solution accommodates 50+ evaporators, 7 trays, and 1 type of black foam, supporting quick addition of new models. Engineers can independently debug and expand without frequent line changes or teaching, truly adapting to multi-variety mixed-line production needs.