Industrial pick & place applications automate processes where loading and/or unloading unitary product is needed. This kind of installations usually use robots or articulated arms to manipulate product from one area to another in stacking, box fitting and palletizing work. Both physical product characteristics (shape, deformity degree and adhesiveness) and disposition, as well as transfer velocity and manipulation zone distinctive features establish task complexity.
To optimize installation efficiency and reduce cycle time, it is common using line tracking technique to take and unload product from or to a conveyor belt while it is running. This is possible with robots being commanded by an encoder signal. Another important point is related to grip design. In some cases, it is enough using simple vacuum grips provided with suction pads which allow to grab piece at desirable speed. A more complex option consists on working with grabbing clamps, pneumatic cylinders or detection devices like photocells and inductive sensors, which confirm presence, correct disposition or piece type placed in the grip.
Cable robot designed by Tecnalia working on a pick&place application
In some cases, product flow disposition is not constant in the process and it may be random positioned over the feeding belt. This problem is usually solved by using a vision system to determine position and orientation of the objects, sending this information to robots by means of an industrial standard communication protocol. These vision systems may act as quality control devices as well, rejecting product in case it does not fit measure, color or shape features previously established. Choosing a suitable vision system depends on many factors, but we can establish some guidelines:
- In case products are presented in random and messy manner over the conveyor belt, with a parallel disposition of them being feasible (a common food industry characteristic, like bakery, biscuits etc.), the most suitable option is to work with a linear camera. This kind of systems allow to configure image dimension to suit it to the product size: tailing consecutive images is usually performed to obtain a belt continuous representation. This allows to save complementary material like presence photocells. Besides, this kind of cameras have some other advantages. We achieve to simplify illumination area, reducing it to a narrow light zone.
- If the product comes one by one in a row and it is suitably separated between each other, the simplest option consists on mounting a photocell which detects product presence to trigger a matrix camera. As a major advantage, this configuration simplifies processing, since we do not struggle with image tailing. The main issue related to this kind of cameras is focused on illumination. In case product exceeds a not particularly large dimension, it may turn to be complex to illuminate a square area uniformly, which may affect to product segmentation.
- Using a 3D camera may be necessary in case product is irregular and it must be taken at different heights. 3D cameras are mainly divided in two types: laser triangulation cameras, which work projecting a laser plane over product to determine profile as it moves along the way and structured light cameras, which work by projecting a known pattern over a rectangular area. The main disadvantage about this kind of devices is related to processing time. Nowadays, it is not possible to apply this technique to high speed continuous flow production plants.
The growing evolution in robotics arms design, which allow quicker and more precise movements, and the increasing resolution and acquiring speed of machine vision cameras, make pick&place solutions a neccesary option to feed the increasing production demand in food industry. In Tecnalia, specifically robotics solutions (cable robot) and image processing algorithms have been developed (pick&pack project) to perform detection, clasification and manipulation of fresh and processed food product.