Vision-Based Robotics Offers Higher Speeds, Precision, and Quality
by Michelle Chaput, DALSA Corporation - Teledyne DALSA Posted 04/27/2005
An increasing number of robotics applications require machine vision for guiding robot movement in automated assembly as well as for quality control. Machine vision replaces human vision with video cameras and specialized computers, and can improve on human vision where precise and repeatable visual measurements and inspections are required.
For automated assembly, components are selected and placed on an assembly by a robot or motion mechanism. A blind robot must have the components and assembly precisely positioned to be able to find and place the components. This requires expensive and often unique fixtures to hold each component type and assembly. A robot with vision can use less costly and more general fixtures and is taught to find and place the components on the assembly. Visual guidance can also compensate for some variations in the components and the assembly, permitting the completion of tasks impossible with blind placement. The additional cost of a vision system for assembly is recovered from improvements in manufacturing flexibility and quality.
An example of automated assembly that requires machine vision is the placement of fine-pitch surface mount components on printed circuit boards. The vision system determines the precise location of the PCB using reference marks (often called fiducials) on the circuit board. The robot picks up each component and holds it in front of the vision system's camera. The vision system verifies that the component's leads are correctly positioned and determines the component's precise location in X, Y, and theta (rotation). From this information the robot arm precisely places the components on the PCB. Typical placement accuracy is better than 1 mil (25 microns) over a 16’‘ x 16’‘ (406 x 406 mm) area.
One of the world's premiere developers of industrial robots is Adept Technology, Inc. Adept designs and manufacturers robots for small-parts material handling, assembly, packaging, and test applications. Products include SCARA robots, linear modules, Cartesian robots, flexible feeders, and vision and motion controllers. Located in San Jose, California, the company has installed more than 15,000 robotic systems worldwide, for such companies as Siemens, IBM and General Motors.
As described above, visual inspection is a crucial part of the manufacturing process. In these instances, Adept installs machine vision controllers and software, such as the AdeptVision VXL, for their customers.
Enhanced Vision Interface
The AdeptVision VXL works as a stand-alone vision inspection system or as an add-on to one of Adept's robot systems. The system is packaged as an industrial cell controller, able to interface to various sensors and other factory systems via built-in digital I/O, Ethernet, and serial I/O ports. The base system is a machine vision controller (EVI), a graphics controller, a motion control subsystem, and the company's ObjectFinder vision software.
The system's Enhanced Vision Interface (EVI) is an OEM version of DALSA's IM-VME high-performance pipeline vision processor. The IM-VME, also available in a PCI version, is designed to excel in high speed, high resolution applications, such as robotics.
The Vision System
The vision controller is a major component of a machine vision system, such as the AdeptVision VXL. The vision system acquires (digitizes) the video information from a camera, processes that information to measure or inspect objects, communicates with the system and motion control, and can display images and information for human viewing.
Extracting visual information requires many computational operations within a short time, so vision computers often use parallel processing to get the necessary speed. The vision computer might also receive input from encoders and position sensors, as well as control the camera exposure and lighting of the object under inspection.
The vision system must extract the information required from a digital image within a specified time and cost. For simple positioning and inspection tasks, a general purpose frame grabber can be used for high speed measurement and inspection—a special purpose vision system may be required. Thus arose DALSA's Modular Vision Computer (MVC), designed for flexibility, speed, and ease-of-use in machine vision and robotics applications.
The IM-VME is driven by a 40MHz pipeline architecture. A pipeline is similar to an assembly line, where raw material enters at one end and a sequence of operations is performed to output finished goods. In a pipelined processor, one or more streams of data (pixels) flow through a sequence of processors (the MVC computational modules), each of which transforms the data or extracts information.
Pipelined processing is a highly efficient processing method where it is possible to perform multiple processing steps for each system clock cycle. In this way the effective computational power is the pipeline clock rate multiplied by the number of processing steps incorporated into the pipeline. Additionally, it is possible to have multiple pipelines operate concurrently. Because of its fast clock rate and advanced computational capabilities, the IM-VME is an extremely fast vision processor able to process a 512 x 512 image through any number of algorithms in less than 7.5ms.
Adept has applied machine vision to a variety of applications, including electronics, telecommunications, pharmaceutical, and food processing. One such application is the manufacture of cookies. In this installation, an automated process spreads chocolate on one half of a cookie and places it on a conveyor belt. Because the orientation and position of the half cookies is random, vision must be used to find them and place the top half. This was previously done by human labor, but quality was poor and people suffered repetitive stress injuries. Adept installed a robot system and the AdeptVision controller and now the robot system finds each cookie half and quickly and accurately places on it the top half.
Adept has also installed its AdeptVision system at Murr Electronik, a manufacturer of electronic products. Murr needed to increase throughput and time-to-market as well as control manufacturing expenses with lower labor costs and no layoffs. Adept installed a robot-based workcell that included one of their SCARA robots, a machine vision controller, the AdeptVision system and V+ software. The system finds each component, picks it up and positions it accurately on the PCB. In addition, a bowl feeder was installed to feed components into the robot's gripper, allowing up to three components to be picked and placed in the correct configuration.
Robots and Vision
Machine vision systems are replacing human vision for quality control inspection of manufactured or natural objects because these inspections are often too fast or precise for human vision. In addition, government regulations and the demand for quality require repeatable vistual inspection on each object. These requirements can be met with a high speed machine vision system integration into the flow of the manufacturing process, either as part of robot guidance or at a station specifically designed for inspection. In either case, the manufacturer wins, producing higher quality products with less waste and, therefore, lower costs.