A groundbreaking collaboration between Fizyr, Cognibotics, and Zivid has resulted in what the partners claim is the world’s fastest automated bin picking solution. The system combines Zivid’s 2+ R-series 3D camera with Fizyr’s AI picking algorithms and Cognibotics’ ultra-fast Honeycomb robot.
System Architecture
- 3D Perception: Zivid 2+ R-series captures true-to-reality point clouds in under 150 milliseconds, even with shiny and transparent objects
- AI Pick Planning: Fizyr’s deep learning models analyze point clouds and determine optimal grasp poses in real-time
- Robotic Execution: Cognibotics’ Honeycomb robot achieves pick-and-place cycle times previously impossible with traditional serial robots
Performance Metrics
The complete system demonstrated sustained picking rates that significantly exceed previous benchmarks:
- Point cloud acquisition in under 150ms
- AI inference time for grasp detection under 50ms
- Total pick cycle time that sets new industry records
- Reliable handling of challenging materials: polished metal, clear plastics, and overlapping items
Why the Zivid 2+ R-Series?
The R-series camera was specifically designed for robotic pick-and-place applications. Its structured-light approach produces exceptionally clean point clouds with minimal noise and artifacts, which is critical for AI-based grasp detection.
This milestone demonstrates that structured-light 3D cameras, when combined with modern AI, can achieve throughput levels that were previously only theoretical. For manufacturers, this means faster ROI on automation investments and greater flexibility in handling diverse product mixes.
Leave a Reply