Bin picking 

Random bin picking is typically at the input stage of a manufacturing process. Instead of traditional static fixtures or pre-filled stacking patterns, a robot is instead emptying a bin bulk-filled with parts for placing on, for example, a feeder, conveyor, or sorter for further processing in the plant. The choice of the vision system and, in particular, the 3D camera impacts the robot’s ability to successfully detect, pick, and place all types of parts.

Zivid Bin Picking
Zivid Two delivers high resolution and extreme precision, native color 3D point clouds with minimal occlusion, and excellent artifact suppression. Significantly improving object recognition and increasing the number of
detectable parts.
Zivid Two enables more reliable detection of object boundaries for grasp planning, and a true to reality representation of object size, rotation, and absolute position in relation to the robot coordinate system. Essential for accurate picking, avoiding mispicks and crashes.
Zivid Two with its groundbreaking trueness, enables more demanding place operations. Dimensioning of objects, and placing for packaging with known position and orientation, with minimal gaps and tight fit, and without colliding or damaging the objects.
Work faster. Zivid Two delivers significantly higher quality at faster speeds than typical laser scanners used in bin picking today. Capture speeds <300ms for high dynamic range scenes, including shiny and reflective objects.
Challenges & Solutions

Bin picking is still a challenge for robots. Even at large manufacturers, adoption of bin picking stations is low, and at SMEs the number is still close to zero. Unpickable objects are typically complex-shaped, hard to manipulate, entangled, and/or highly reflective machined or polished metals. The difficulty stems from the inherent uncertainty in physics, perception, and control.

See tiny and detailed objects, densely stacked or randomly arranged.
Zivid Two provides high resolution and precision point clouds with minimal occlusion, capturing shapes and sizes with all the fine details.

  • Very small, thin sheets, highly detailed, and irregular objects.
  • Densely stacked with a little gap or randomly arranged and piled on top of each other.
  • Distinguish features smaller than 5mm for reliable object recognition and separation of boundaries for picking.

See shiny, reflective, machined, and polished parts

3D HDR and Artifact Reduction Technology (ART) ensures excellent suppression of imaging artifacts from reflections, interreflections, specular highlights, and high contrast transitions.

  • Shiny and reflective, sheet metal, cylinders, washers, molded, machined, polished, and chrome-plated parts.
  • Capture speeds <300ms for high dynamic range scenes such as shiny and reflective

Experience fewer occlusion outliers 

Zivid Two is ultra-compact with a small baseline and more optimal occlusion performance than larger baseline cameras and scanners. A large baseline gives excessive projector/camera shadowing and loss of scene details, creating holes, and missing data affecting the detection. There's also a risk of missing smaller objects hidden by shadows in the bin, for example, close to corners and edges, especially when picking from smaller boxes.

Mech-Mind Bin Picking
Guided along by the powerful 3D vision system, the robots can recognize randomly-piled materials, even those with dark or reflective surfaces and complex structures, and then pick them up from deep bins accurately without damaging the components. 
Core Advantages
High Cost-Effectiveness

The price is only half to one-third of the same type of typical products.

Powerful Algorithms

Built-in AI algorithms enable robots quickly and accurately recognize objects of random sizes and shapes (including cartons, boxes, bags, etc.) without registration.

Intelligent Path Planning

The built-in path planning algorithm guarantees the movement of robots is precise and collision-free.

Plug and Play, Easy to Use

A visualized interface can simulate the robots' movement with only one click.

The code-free programming enables a low threshold for operators to deploy.
Seamless Integration with the WMS System

Seamless integration with the WMS system allows robots to pick up goods on demand.

SpeedSingle cycle time can reach 3s
ObjectsSupport objects in different sizes and shapes (including randomly stacked tiny objects, complex-structured metal parts with dark or shiny surfaces, etc.)
Robot BrandsCan be adapted to various mainstream robot brands, such as ABB, KUKA, YASKAWA, Kawasaki, Rokae, Peitian, Techman, Estun, etc.

Personalized solution;

Staff training;

Fixture design.

Intelligent Planning Algorithm
Mech-Vision Graphical Machine Vision Software
Mech-Vision simplifies all codes into steps, and users can edit algorithms without writing any codes.
It can support users to develop customized algorithms and independently deploy multiple typical applications.
Mech-Viz Interface

One-click simulation on the left side of the software. Users can intuitively see the robot's motion path, recognition results and collision model.

The right side is the robot operation process, which is easy to learn to operate.

Different from traditional code-based programming software, Mech-Viz is a graphical and visualized programming environment.

Built-in path planning, collision detection, and grasping strategy, and other intelligent algorithms help the robot plan the optimal route and avoid collision.

With Mech-Viz, users can manipulate robots after simple training.

Our solutions can be adapted to various mainstream robot brands.

Camera Introduction

Point Cloud
Randomly-Placed Metal Parts
(Track Links)
Randomly-Piled Steel Plate with Various Specifications
Randomly-Placed Metal Parts (Crankshafts)
Powered by
This website uses cookies for best user experience, to find out more you can go to our Privacy Policy  and  Cookies Policy