The Emergence of Bin Picking
Bin picking is a robotic material handling task that involves implementing a 3D machine vision camera unit with image analysis algorithms to allow a robot to pick-up randomly located and orientated parts from a bulk storage bin. These parts can then be placed in precise locations ready for the next stage of their automated journey. In this way parts can be singulated and orientated for the next process efficiently and accurately.
Random bin picking plays a pivotal role in advanced manufacturing environments and is suitable for applications that have tight cycle times and require accurate placement of parts from large storage containers.
What does a typical bin picking setup include?
1. A bin or container full of pickable parts
2. Machine vision unit
3. Industrial PC with machine vision software
4. Robot / Cobot + controller
5. Custom robot gripper: magnetic/ vacuum/ mechanical
The main brain of the whole system is the machine vision unit. It comprises of a light projector and 2 or more camera sensors. The light projection can either be a structured light source or a laser scanner. Structured light systems project a known 2D light pattern over the bin; this could be a point cloud or similar dual axis pattern. Laser scanning systems sweep a laser line over the pick area allowing laser triangulation which builds the 3D surface slice by slice as it sweeps over.
What are they advantages of bin picking?
Bin picking can reduce costs and improve productivity by achieving better cycle times.
The application of bin picking reduces double handling of parts. It also allows for great density of product storage, simplified factory logistics and reduced floor space.
Less labour is required to feed the line overall and it saves humans from having to do highly repetitive, monotonous tasks. This creates opportunities to increase production and reduce the margin of error in operation.
The system can operate inside the guard lines of other machinery.
Bin picking opens the possibility to design the production line with further automation like Automated Guided Vehicles (AGVs) and Autonomous Mobile Robots (AMRs)
What are some challenges of bin picking applications and how can they be overcome?
Tangled or nested parts in a bulk bin can be a challenge. The effects of this can be mitigated by utilising the algorithms to choose to pick the most accessible part in the whole bin each time. Although most of the time there is an option to pick up a part for smooth operations, the system can be programmed to pick up tangled or nesting parts all at once and drop them from a given height to hopefully aid in untangling them and reveal a part free of obstruction.
Orientation – depending on the geometry of the part, sometimes 3D localisation algorithms are not possible. To overcome this an additional 2D camera and/or mechanical jigs or squaring tables can be used to figure out the correct orientation of the part.
Ideally systems can be used with ambient lighting conditions which keep costs down and provide flexibility. However sometimes inconsistent lighting conditions can cause complications, particularly with highly reflective product materials (which are typical in our industry with sheet metal brackets etc). This can be addressed by implementing a controlled lighting environment. This could be in the form of a shroud which eliminates the variability of overhead lighting. Taking this a step further, you can fully encapsulate the whole system in a cabinet and have controlled lighting within. This will allow you to calibrate the vision system once with consistent lighting parameters.
What are some of the technological advancements that provided easier solutions to manufacturers?
A lot of robot companies have started directly collaborating with other companies and/or are developing their own machine vision systems and software to create “off the shelf” solutions for bin picking which can be adapted to specific applications as needed.
Some CAD matching algorithms have been developed by multiple players which can locate and match 3D CAD files in the bin picking zone allowing the exact 3D orientation of the parts to be known within the space. This allows even more optimisation of target selection, pick location and path planning.
Robot arms are becoming more and more sophisticated where they have more slimline design, smaller footprint, increased work envelopes and higher payloads. Cobots have developed greater capabilities by way of increasing their reach and load ratings. This has closed the gap in scope of suitable applications between them and their industrial robot counterparts.
A more recent technological advancement in bin picking is on board processing and application-specific software modules running on the camera. This standalone system can run the vision and robot picking software off its own processor which is built into the unit so an external PC isn’t required.
How do you simplify bin picking applications?
Simpler solutions often provide the robustness needed to achieve success.
Using a gripper designed specifically for the application is worth the investment. This makes the system reliable. The functionality of the gripper is what makes or breaks a bin picking solution, because if you can’t achieve a high success rate of picking it doesn’t matter how good your vision system is.
Adding an additional 2D camera to the system or a mechanical jig can add a layer to the process which will allow the system to account for all possible scenarios. For example, it may allow the robot to pick up a part from the bin on the opposite face than desired then place into a mechanical jig which “corrects” or aligns the orientation ready to be picked up again in the correct orientation. This side process may not be needed for most picks but can be utilised on as needed.
The application and shape of the part will determine how complex an algorithm is needed to locate the part from the images. If a more basic algorithm can meet the requirements of the application significant saving can be made by selecting a lower spec’d vision hardware.
Cobots vs Robots
Many standalone bin picking systems incorporate Cobots which are great for flexibility; however for most industrial automation projects here at Facteon, we typically use industrial robots because they provide more efficient, robust and cost effective solutions.
Main deciding factors:
- Industrial robots will typically offer higher loads, longer reaches and faster speeds at the same price point
- Cobots can operated in conjunction with safety scanners with reduced speed safety zones
- Cobots can be taught by hand: requires less skilled operators to train/retrain the robot
Find out more:
Speak with one of our team to find out whether your business will benefit from vision technology. Contact Facteon here via our contact page.