Picking the unknown with advanced vision AI – Fyzir
This video explores how Fizyr’s advanced computer vision software enables robots to pick completely unknown items in complex logistics environments. You’ll learn how the system uses segmentation, classification and depth of field grasp‑pose estimation to deliver highly reliable results – even with mixed parcels, reflective packaging, or deformable items.
The video explains why automation in parcel handling is increasingly critical, how millions of images were used to train a robust neural network, and how technologies such as IDS Ensenso 3D cameras and uEye RGB (RedGreenBlue) sensors integrate seamlessly into the workflow. You’ll see practical examples from logistics, groceries, apparel, depalletising, and truck unloading, demonstrating how the solution generalises like a human while maintaining industrial‑grade precision.
Topics covered
- How segmentation prevents mispicks in dense or complex parcel layouts
- Classification of item types for correct grasping and quality control
- Use of 3D point clouds and grasp‑pose estimation in six degrees of freedom
- Practical examples from logistics, depalletising, apparel, and groceries
- Benefits of supervised learning and hardware‑independent system design
Video timeline
- 00:00 - Introduction to Fizyr
- 01:00 - Why Fizyr is in logistics
- 02:40 - The solution of Fizyr
- 03:02 - Picking use cases
- 05:51 - Fyzir use case: finding the corners of towels
- 07:50 - Pick and place use case from automatica 2018
- 08:50 - More Vision applications
- 11:50 - Market Opportunity
- 12:17 - Technology of Fizyr
- 13:45 - Segmentation examples
- 16:25 - Grasp-pose examples
- 16:53 - Use case: black and reflective parcels
- 17:58 - Q&A
Who should watch
Ideal for robotics engineers, system integrators, logistics professionals, warehouse automation designers, and anyone evaluating AI‑powered item‑picking systems.
Watch the full video to discover how intelligent vision enables robots to pick the unknown cutting‑edge automation.
