Bin picking is one of the most challenging tasks in industrial automation. When parts are stacked, overlapping, reflective, or partially hidden, it becomes difficult for robotic systems to identify, locate, and pick them reliably under real production conditions. AI-MATTERS helps manufacturers and AI developers address this challenge through a realistic testing and experimentation service for AI-based product bin picking, combining 3D vision, robots and cobots, simulation tools, real industrial parts, and validation services.
The Challenge
Bin picking may seem straightforward, but in practice it is a highly variable task that combines perception, motion planning, gripping, and cycle-time performance. Parts can be occluded, randomly oriented, or difficult to detect because of their shape or surface properties.
To manage this, manufacturers have often relied on feeders, trays, fixtures, or manual pre-alignment. These solutions can simplify handling, but they also increase cost, reduce flexibility, and take up valuable space. AI-based bin picking offers a more adaptable alternative by reducing the need for custom jigs, manual sorting, and operator intervention.
The challenge is also operational. A poor pick can interrupt machine loading, assembly, inspection, or transfer processes.
The Solution
AI-MATTERS provides a complete service for testing and validating AI for product bin picking in realistic manufacturing scenarios. The goal is to verify that the full system works reliably with real parts, vision and robotic constraints, and production variability.
At the core is an advanced 3D vision-guided robotics system that captures detailed 3D data using a four-camera, one-projector setup. Multiple striped-light patterns are projected onto the target, generating 136 images that reconstruct the scene with high precision. This supports accurate detection of randomly placed parts and reliable 3D pose estimation.
The setup process is designed for fast deployment, with automatic robot-camera calibration and the option to upload CAD data for the target part. This shortens preparation time and simplifies configuration for new products or variants.
This service also provides access to industrial robots and cobots, allowing users to test the interaction between vision, robot motion, gripping strategy, and cell design. The built-in Picking Simulator adds further value by allowing different grippers and layouts to be tested virtually before making physical changes, reducing cost and speeding up iteration.
Validation with real parts, including components from sectors such as automotive and white goods, makes the testing more meaningful. The service is further supported by a data acquisition system and APIs, enabling users to monitor throughput, failed picks, repeatability, and disturbances, and to refine the system based on evidence.
AI-MATTERS also offers a collaborative experimentation environment where manufacturers, technology providers, and AI developers can work together to optimize the solution for specific industrial needs.
How we help you
- Test AI-based bin picking in a realistic industrial environment
- Validate perception, pose estimation, and picking logic on real parts
- Experiment with vision systems, robots, cobots, grippers, and cell layouts
- Use simulation to assess reachability, collisions, and picking performance before hardware changes
- Access open infrastructure (e.g data acquisitions, APIs) to analyze results and refine your solution
- Reduce technical risk before deployment at customer sites or in production
Ready to take the next step toward industrial deployment?
If you are developing an AI-based bin-picking solution, or if you are a manufacturing company looking to integrate intelligent robotic picking into your production processes, AI-MATTERS offers a practical environment to test and validate the technology. With access to industrial equipment, real parts, simulation, and testing expertise, you can refine performance, reduce risk, and move forward with greater confidence toward industrial readiness.