- Home » News » Product News
Vision sensors help robots to pick parts from belts or bins
At the recent Automatica exhibition in Germany, the German sensor-maker Sick showcased a series of 2D and 3D vision-guided parts-locating technologies for quick and easy set-up of robots picking items from a belts or bins. Various sensing options are available – including streaming cameras, vision sensors, and smart cameras – offering a choice of measuring ranges, fields of view, and resolutions to suit particular tasks.
Sick was demonstrating three systems designed for affordable, entry-level automation and handling, especially of small parts and components:
• the PLR robot guidance system is designed to locate parts in load carriers – providing visual guidance for a robot when it is removing car body panels from a transport frame, for example;
• the PLOC2D is an easy set-up vision system for 2D location of parts or products to be picked from a static workstation, moving belt, or feeder system; while
• the PLB 520 uses a stereoscopic camera to provide 3D vision-guided bin-picking of much smaller objects than was possible previously.
These systems are said to offer good precision, repeatability and reliability, and short cycle times. They can be taught a wide range of parts either directly or via a CAD system, and they incorporate software tools to measure parts and communicate with robot controllers.
The PLR guidance system uses a laser sensor mounted on a robot arm to sense the position of a load carrier and the parts it contains. It provides the robot controller with precise 3D coordinates for removing parts. It can compensate for the carrier being placed imprecisely, parts moving in transit, or dimensional deviations resulting from production. It is not affected by reflections from components or by other environmental factors.
The PLB (Part Localisation for Bin picking) system is aimed at applications where a robot needs to remove pre-positioned, unsorted, or stacked parts from a lattice box or a container for loading into machines. It consists of a high-resolution 3D stereo camera, software for 3D shape comparison and parts-location, as well as measuring and communication tools for robot integration. New parts can optionally be programmed via CAD and transferred to a robot-assisted part-handling system. Reflections from metal do not affect the results or accuracy. The PLB can also help to avoid collisions with the tote when positioning the gripper to remove parts.
The latest PLB 520 model consists of two 1.3-Mpixel cameras, with high-power blue LED illumination for good rendition of a wide range of object finishes. The simple teach-in procedure uses uploaded CAD models of parts, together with several identified gripper positions, to enable the algorithm to locate the part in any orientation and select the most appropriate gripper position to pick up the part.
“Previously, most 3D part localisation systems were designed for larger-scale applications,” explains Neil Sandhu, Sick’s product manager for imaging, measurement and ranging in the UK. “Now typical tasks like picking specific small parts – such as bolts from a deep mixed parts bin – and placing them on a conveyor, or selecting part-completed items and placing them on a press or machining centre, have become more feasible for smaller operations, such as machining or assembly sub-contractors.”
The PLOC2D system provides automated 2D object location for continuous picking of components and parts from a static workstation, moving belt or bowl feeder. It includes a 2D camera with built-in illumination, and optional optics with fixed or flexible focus, and supports single or multiple detection of parts. It can identify one specific part type out of multiple types it has been taught. The sensor software includes motion synchronisation. A conveyor-tracking function can provide coordinates to a robot to grasp moving parts from a conveyor belt. A Web-based HMI and a user-friendly process for teaching a variety of parts, result in short set-up times. The PLOC2D has a 2 x 2m field of view and can locate more than 120 items per minute. The PLOC2D’s EasyTeach function matches taught shapes against the object shape measured down to a 0.5 pixel resolution in two axes.
Both the PLOC2D and PLB 520 can be connected directly to a robot controller without programming skills or training and are ready for use almost immediately. According to Sandhu, they will facilitate two of the most common robot tasks. “Especially when combined with a cobot system, they provide new opportunities for production teams to replace repetitive or physically-demanding manual tasks with a more affordable robotic solution that is straightforward to set up.
“These solutions are the first for Sick,” Sandhu adds, “using its new AppSpace software development platform, to deliver easy-to-use, vision-guided part localisation for robotics. They open up new applications for robotics so production engineers and machine designers alike can use cobots flexibly for different uses across a production facility, and even test out different automated processes. This sort of robotics is often not a high-speed substitute for manual picking, but replaces a human’s repetitive and mundane task with a safer, high-consistency alternative.”
