Depth sensor

Overview
AI-based picking solution for industrial automation. The 3D Handling System enables automated 3D bin picking even with varying object orientations, challenging surfaces and without in-depth image-processing knowledge.

Key product points
Automate demanding bin-picking processes with AI-based 3D Vision.

• 3D bin picking with the proven VISOR® cameras
• Reliable detection of challenging surfaces and complex shapes
• Quick start with cloud-based AI training; no specialist image-processing knowledge required

Advantages at a glance

• Automation of manual gripping and positioning tasks: Replaces previously manual steps to increase production efficiency.
• Reliable detection of demanding objects: Handles transparent, glossy or otherwise difficult surfaces and complex shapes with process reliability.
• Easy setup with no specialist knowledge: Upload a CAD file, configure the application and start without extensive parameter configuration.
• Greater process reliability: Integrated collision avoidance reduces the risk of damage to robot, grippers and system.
• Stable workflows with less downtime: Automated processes increase shift capacity and deliver reproducible results.
• Faster ROI: Higher quality, shorter cycle times and 24/7 operation support quicker amortization.

What is 3D bin picking?
3D bin picking enables robots to pick randomly arranged parts from bins, boxes or other containers and position them for the next process step. In the 3D Handling System an AI-based vision system provides the 3D data required for reliable gripping and positioning. The solution pairs field-proven VISOR® cameras with AI evaluation software and a local computing unit to capture image data and interpret it spatially for precise position information.

Included in the 3D bin picking vision solution

• Computing unit with pre-installed software: Industrial PC with NVIDIA GPU for AI-based part detection and robot communication.
• Camera module including bracket: VISOR® industrial camera system with integrated illumination and auto focus for reliable image capture and easy integration into a robotic cell.
• Accessories and calibration: Calibration plate and mounting/connection accessories for fast start-up.

Suitable applications
Designed for use cases where traditional image processing reaches its limits: unorganized bulk materials, irregular shapes, reflective or transparent components and changing illumination. The system enables automation of tasks that previously required manual handling or extensive integration effort.

Compatibility
Compatible with leading robot manufacturers and controllers for direct integration and robot communication.

Implementation workflow (5 steps)

• AI training: Cloud-based AI training using CAD models.
• Image capture: Two VISOR® cameras with integrated illumination and auto focus capture the scene.
• Runtime analysis: Local processing on a Linux PC with a high-performance NVIDIA GPU.
• 3D position determination: System calculates 3D position and orientation of components.
• Robot communication: Output is transferred directly to the robot controller to effect gripping.

Technical specifications

• AI-based 3D vision for reliable part detection and pose estimation.
• Two VISOR® industrial cameras with integrated illumination and automatic focus.
• Computing unit: industrial Linux PC with high-performance NVIDIA GPU (pre-installed software for AI evaluation and robot communication).
• Includes camera modules, mounting bracket, calibration plate and connection accessories.
• Cloud-based CAD model AI training and local runtime analysis for low-latency operation.
• Provides 3D pose output for robot controllers; integrated collision-avoidance support to protect robot and tooling.
• Designed to handle transparent, reflective, glossy and irregular-shaped components.
• Compatible with major robot brands and controllers for seamless integration.