Intelligent Bin Picking

Model-free grasping of unknown objects with low-cost camera and ROS

Benefits

Cost Effective and Flexible

Details Avoids complex, inflexible and expensive feeding systems. The software works with the low-cost RealSense camera.

Reliable Gripping

Details Monitors the gripping process to ensure an accurate, rapid and secure grip.

Plug & Play with ROS interface

Details Allows an easy plug & play setup due to a tight integration of the camera, software and gripper with standardized ROS interfaces.

Versatile bin picking

Details Recognizes and grasps a large variety of objects using deep learning technology. Object models or pre-training are not required!

Inquiry

Get in contact for more information or for a free analysis of your picking use-case!

Contact us

Packages

Our bin picking solution is available in various packages, ranging from software only to a fully integrated solution, which includes services and hardware, such as robot arm, gripper and camera. Select the following modules as required:

  • Bin picking software

  • Interface & Web UI

  • Tactile sensitive grasping

  • Intelligent gripper & camera

  • Robot arm

  • System design & Integration

  • Deep Learning

  • Bin picking software: The core of our solution, an intelligent AI-based bin grasp planning software is available as a stand-alone software module. Thanks to its standardized Robot Operating System (ROS) interfaces, integration with any ROS-enabled robot arm or gripper is straight-forward.
  • Intelligent gripper: Packaged with ViseGrip, we can provide a complete EOAT solution for intelligent bin-picking. ViseGrip integrates an industrial gripper, an advanced motor controller and a stereo vision system. It also features camera-based sensors for tactile data (ViseTac) for online grasp control and sensitive gripping. Learn more about ViseGrip, Camera-based tactile sensors (ViseTac)
  • Full system: A full bin-picking system has been developed jointly with KEBA Group, integrating KeMotion Flex Core and arms from Kawasaki, Stäubli, or others. For the entire system, communication relies on standardized open ROS interfaces, allowing for custom extensions by the client.

Specification

Component Features
Camera
  • Intel RealSense low-cost depth camera
  • Compatible with most other depth cameras
Compatible robot arms UR, Kawasaki, Stäubli, any ROS-enabled arm
End-effector
Sensing and feedback Camera-based tactile sensing ViseTac for:
  • Sensitive grasping for soft and fragile objects
  • Automatic grasp force adjustment
  • Grasp verification
Suitable objects
  • Broad range of objects, typ. 1cm to 10cm
  • Especially fragile/soft objects, e.g. glass, plastics, fruits/food
  • Graspable by two-finger gripper
Grasping algorithm Model-free AI-based grasp point detection and scoring. Grasps unknown objects, no training/object model required.
Interfaces
  • ROS (for camera, arm, etc.)
  • ROS fully exposed or hidden in container, as desired
  • Web interface
  • KEBA KeMotion Flex Core
Detection time Typ. 1s detection, instant scanning
Processor Standard PC, mid-range/high-end i7 recommended. Installation on existing PC possible.

Contact

Visevi Robotics GmbH
in WERK1, Atelierstr. 29
81671 Munich, Germany