Bin picking in response to rising volumes in pharma industry
Exploding demand for diagnostic and test kits for COVID-19 means pharma companies need to shift up a gear. The right vision system is crucial for bin picking in an industry that is all about extreme precision and delicate parts. In this article, Pickit shares their insights and best practices.
Automatically picking parts from trays or containers and then offering them for further processing has the greatest potential in terms of automation.
Indeed, in the pharma industry, it was previously difficult to reliably automate this process. Because it often involves delicate parts. Those parts are typically taken out of a box by an operator and then delivered to the robot or machine in a certain way.
In manufacturing plants, mainly in heavy industry, a similar challenge does have automation. There, randomly scattered parts in a bin are automatically prepared for the next step in the process by giving the robot eyes. Vision technology, in other words. Pickit controls the robots to pick parts directly from a bin, tray, box, pallet or conveyor belt. To then deliver them correctly to the machine.
Switching is done at the touch of a button. Instead of taking up physical space to stock various jigs, clamps and feeding devices, the vision algorithms are on a PC.
Bin picking is thus native to heavy industry.
Not so in pharma.
Just for the reasons already cited: not accurate enough, not performing enough. Using 3D vision technology for pick-and-place robots in the pharma industry? Not common.
However, recent projects prove the success of automatic detraying (or retraying) of pharmaceutical products using a robot. An important sub-process in drug manufacturing, is the separation of materials for further processing. Known as detraying or retraying, this process is fine for tablet blisters or rectangular pharmaceutical packaging. But the story becomes more complex when it comes to transparent syringes, which are typically delivered as bulk goods in containers or boxes. In that case, it is often workers who manually transfer the syringes to trays.
For this application, however, our vision system was used as one of the building blocks for a more scalable bin picking system. For this case, that resulted in:
- short cycle times,
- high quality standards,
- and flexible and easy retooling.
Another application involving robots with vision technology was manipulating medicine packaging. This involved robots picking the packages from larger containers and placing them in the storage system.
These applications prove that bin picking, even if new to the pharmaceutical industry, has plenty of advantages. For both clinical production (for R&D purposes) and commercial production (for patients).
How does it work?
Noise-free images
The HD camera with structured illumination creates noise-free images.
This is how we eliminate the ambient light factor. Structured light is the process of projecting a known pattern (often grids or horizontal bars) onto a view. These grids or bars distort when they fall on a surface. This allows the vision system to calculate the depth and area of objects.
Detecting transparent parts
Detecting transparent parts using vision technology is challenging. And in the pharmaceutical industry, transparent parts are unavoidable. Still, it is possible to identify parts such as syringes. Thanks to labels and caps. That’s all our cameras and algorithms need to identify the whole thing.
Detecting small vulnerable parts
Small parts make bin picking extra complex. That’s true in any industry. Our system generates a 2D image of the bin, and by doing the visualization from above, we bring overview. The system manages to detect the parts so the robot can pick them.
Careful handling of delicate parts
Another critical aspect of bin picking is that the system must be able to remember previous operations to avoid collisions. This is crucial in the pharmaceutical industry, as it often involves delicate parts. Our systems provide collision-free bin picking by using modeling tool technology.
This technology allows us to model the robot tool. The model of the robot tool is useful to visually confirm the correct location of a collection point. In addition, it can also be used to prevent collisions between the tool and a bin or other objects.A collection point indicates where the robot can pick up an object.
It is specified as a position and orientation relative to the object, where the robot Tool Center Point (TCP) must move in order to pick up an object. A good pick point depends on both the gripping tool, and the object to be picked. In scenarios with limited picking capabilities such as bin picking, it is possible to avoid collisions between the robot tool and the bin or other objects.
When defining collection points after object detection, it is useful to visualize the robot tool relative to the object.
In order to confirm the correct placement of a pick point. If the application requires it, it is possible to create multiple tool models and assign different tools to different points.
For example, a common example is a gripper with two fingers, combined with different opening distances depending on the gripping point selected.
Bin picking in pharma: conclusion?
The advancement of vision-based technology in bin picking ensures that high-quality production processes can be automated through accurate positioning, online monitoring, and real-time coordination.
The pharmaceutical industry has much to gain from automating bin picking. So do other industries, including aerospace, automotive and food and beverage, among others.
The broadest possible view of mechanical engineering and retrofit is the best guarantee of success. Therefore, in this blog we give the stage to another expert within our broad field.
Pickit emerged in 2016 from Leuven-based Intermodalics, a robotics software developer. Pickit is a 3D vision software for robotic grippers. Automation engineers in more than 40 countries use the software to efficiently bin gripper arms, depalletize or locate assembly parts.