23.17.
Robots with laser probes and machine vision systems are finding more and more applications in flexible inspection. In flexible inspection, sensors are moved about the part being inspected along a programmed path trajectory. Robots with articulated arms can reach inside of parts with cavities, such as car bodies. Robots can be programmed to measure reference datum points and planes on a part. Algorithms, using these reference point measurements, can calibrate the part co-ordinate system, which allows the CAD/CAM database to be integrated into the robotic measuring system. This integration eliminates the need for precision part fixturing, allows real-time adaptive robot path trajectory control, and provide off-line programming of robot path trajectories. Control of robotic inspection systems by powerful minicomputers, allows quick down loading of path trajectory programs to recalibrate a current part co-ordinate system, or to reprogram robots to inspect a new part.
Robots equipped with optical and vision sensors provide the maximum flexibility to carry out a wide range of inspection tasks. While optical sensors provide accurate, non-contact measurement of distances and ability to accurately measure co-ordinates on plane and curved
surface that would require structured lighting for vision measurements, the vision systems allow the inspection of part features, having a variety of geometrical shapes.
Inspection robots are presently used for moderate through-put rates. Robot positioning accuracy is the main source of error in robots inspection systems.
Selection of Robot Based Inspection System
Robot size. The robot size is determined by the inspection volume required by the application. The revolute co-ordinate geometry design or gantry design of robot is widely used for inspection applications. The gantry design, provides high repeatable positioning accuracy due to its geometric design. Robot positioning accuracy is determined by its ability to match its actual position in 3-D space to the command position called for by the program’s position instruction. The error between actual and command positions may be caused by servo component design, structural natural frequencies, bearing friction, gear backlash and load torque. As the robot approaches its command position, the dynamic errors produce an effect called “setting time”.
Sensors. Either vision systems or optical sensors are used. In vision systems, binary images allow the use of simple algorithms for image processing and introduce least delays into the process inspection cycle. With good silhouette lighting and binary image processing, it is possible to obtain reliable measurements to a one pixel accuracy. Since this accuracy is limited by robot positioning accuracy, the pixel calibrations is set equal to 2.3 times, the standard deviation of the robot positioning error distribution.
Optical sensors employ a light beam to illuminate the measurement location and use reflected light and triangulation techniques to measure distance. In order to permit measurements on surfaces with poor reflectivity light frequencies in the near infrared range are used. Detectors used to sense light beam deflection, caused by the trigonometric relationships, are of either the line scan matrix type or a semiconductor film employing analog circuitry. For latter type close-loop control of light beam intensity is necessary to maintain constant reflected light intensity. The measurement accuracy of optical probes is high and these have high speed measurement capabilities.
Computer control systems. A robot inspection system uses 3 or 4 small computers to perform difference distributed processing functions. These computers are interconnected by a data communication and control network, thus permitting a host computer to co-ordinate and control the robot and sensors.
System software. It must provide the following functions :
initiate and co-ordinate the process inspection cycle.
read in and store sensor measurement data during inspection cycle execution.
perform data processing.
output process data results in some acceptable report format.
provide a limited operator interface to handle calibration and diagnostic requirements.
output warning and diagnostic messages, to indicate system malfunction and type of errors.
