Bin Picking Highly Variable Complex Shaped Parts

Context & Aim

Many sectors in industry produce batches of material in random orientation and need a better way to sort product that has large variability, complex non-geometric shaped parts. Presently, considerable in line automation or manual effort is required to sort parts that have large natural variation. Bin picking solutions are available but there are trade-offs in terms of speed, product shape, and accuracy of pick and place. Many systems do not have the capability to deal with large variability in product shape and size. 

Challenge

Research commercial bin picking solutions and build a prototype demonstrator for random bin picking, disclose recommendations giving  consortium members an insight into bin picking solutions for complex-shaped variable products.

Technical challenges include:

  1. Random bin picking of objects with a certain level of variability (±15% of object volume)  
  2. Cycle time maximise, 
  3. Complex non-geometric shaped product. 
  4. Non-symmetrical shaped product 
  5. Alignment on pick and place 

 

The Product chosen to build the system around was croissants that satisfied the requirements for variability, nonsymmetrical, and complex shaped parts.  

Solution

  1. IMR Engineers developed a full bin picking solution to locate and pick variable complexed shaped parts from random bins. The solution was designed and tested in a virtual environment. The final physical build was done with both collaborative and industrial robots, giving a cycle time of 9ppm 
  2. The key elements required to build a bin picking software solution include 3D Vision, Object Identification, collision avoidance, trajectory planning,  Bin design, and End of arm tooling design. 
  3. The System was built using a Kinnex 3d sensor and open source software ROS, Movit, and Halcon vision software from Multipix.
  4. The development of the solution was carried out in 6 Stages:
    • a) Assessment fo product and EOAT design. 
    • b) Build virtual environment– this allows for quick testing and iteration of the system
    • c) Calibration of the robot, vision systems, and product 
    • d) LocalisationLocalising product in all orientations and positions 
    • e) Full System build: Integration of hardware, software, and safety. 
    • f) Optimisation of the system

Results

  1. The system achieved a cycle time of 10ppm comparable with commercial solutions.  
  2. A range of commercial solutions identified, and feasibility work carried out on a small selection 
  3. ROI calculations to aid in business decisions 
  4. Closeout  
  5. Knowledge sharing and insights 

Impact

  1. IMR has built capability and expanded knowledge in the area of vision systems and robotic bin picking 
  2. IMR’s consortium partners have gained an insight into 
    • A. Commercially available solutions 
    • B. The technical complexity of the system an issues 
    • C. Improved knowledge of applications, benefits, and limitations.  

Benefits to the client:  

  1. Demonstrate to REC partners the capability of bin picking  
  2. De-risk the uptake of bin picking applications in Irish manufacturing. 
  3. Develop skills within IMR to address the technological challenges associated with bin picking.  
  4. Demonstration of both the IMR system and commercially available systems. 
  5. Background research report 
  6. Detailed Technical report. 
  7. Information on commercial viability for various scenarios.   
  8. Demonstration a proof of concept system and disclose recommendations allowing consortium members to make an informed business decision on the competitive advantages posed by bin-picking solutions 

PROJECT MANAGER

Kevin Dooley
Senior Robotics Researcher

THEMATIC PILLAR

Automation and Advanced Control

FOLLOW US