by David H. Genest

With the automotive industry's continuing push to reduce inspection cycle times, traditional tactile probe and data-gathering technology is often too slow to support most full-body and body-assembly inspection requirements for timely process-control applications. Consequently, automakers and coordinate measuring machine manufacturers are continually evaluating improved methods for gathering and analyzing dimensional data.

One new approach focuses on high-speed scanning using CMMs that automatically measure the shape and form of workpieces. The CMMs then incorporate the information with CAD systems to provide insight into the manufacturing operation. High-speed scanning is simply a way of automatically collecting a large number of data points to define a part's shape quickly and accurately. Such noncontact gaging--when used for subassembly and body-in-white inspection applications--offers accurate real-time monitoring of quality output from the production line as well as the flexibility to handle line changes.

For several years, CMMs have been used as in-line measurement and inspection systems for body-in-white applications. In many of these cases, however, CMMs are equipped with tactile analog probes to gather dimensional data. These probes are durable, heavy-duty electronic sensors that provide a high degree of data accuracy, albeit at a relatively slow speed.

Scanning technology was initially developed during the mid- to late 1970s. Beginning in the late 1980s, microprocessor-based controllers and firmware were introduced. At about this time, the RS-232 connection between the CMM and host computer changed to Ethernet, and clock rates increased to 200 hertz. Today, more PC-like controllers and 500+ hertz clock speeds have made high-speed data manipulation possible. An Ethernet connection between the CMM and the host computer can transfer information at 100 MB/sec or faster. Because of these advances, as well as the widespread use of finite element and modal analysis in coordinate measuring, CMMs can now gather thousands of data points per second using advanced sensor technology.

Problem solving with noncontact sensors

Corresponding with improvements in CMMs and controller technology are improvements in the field of sensor technology. In particular, two areas of noncontact sensor technology for measuring car bodies and subassemblies have received much attention: laser stripe scanners and optical sensors that use a CCD camera combined with a suitable lighting device.

Both of these sensors integrate a light source and photoelectric detector and work on the principle of triangulation. The light source emits a precisely focused laser or infrared light beam that strikes the workpiece, creating diffused or scattered light that is then focused on a photoelectric array. Any variation on the surface distance from the sensor results in a change in the spot image's position on the array.

The technology has been successfully adapted to measure all typical body-in-white features, such as surface point, center of gravity, character points, edge point, large and small holes, curvature radius, slot, square slot, and flush and gap.

Laser sensors work by projecting a strip of light on a part's surface to make a virtual copy of an existing part or shape. This virtual 3-D copy consists of a cloud of measured points, each with its own XYZ coordinate in space; the point cloud is used to perform inspections. Although laser sensors are highly accurate data-gathering devices, they cover less area during the data-gathering process than CCD-type optical sensors. A typical laser sensor offers a field of view of 20 X 20 mm, a maximum resolution of 40 X 60 µm and accuracy in the range of 15 µm. Laser sensors generally capture 500 data points per line and scan at a speed of 25 lines per second.

The CCD optical sensor works by relating the pixels of the image or picture taken by the sensor to the corresponding 3-D points of the framed part area. It allows a virtually simultaneous reconstruction of all the points inside its field of view. The sensor's scanning speed is directly proportional to its field of view. For example, a typical optical sensor of this type has a field of view of 59 X 42 mm, a standoff distance of 117 mm, a field depth of 8 mm and a resolution of 8 µm/pixel. The CCD optical sensor's broad data-acquisition area allows inspection routines to be performed very quickly. Compared to a tactile probe measuring the same feature, the cycle time is significantly shorter: for example, 1.5 seconds for a noncontact sensor compared with 16 seconds for the tactile probe.

Some noncontact sensors integrate a structured light source that emits a plane of light. When this plane intersects the part, a line of light forms along the part's contour. The image sensor detects this line and transforms it into a measurable digital image. Distance measurements can then be obtained based on the shape and position of the line on the part surface. Structured light sensors are able to simultaneously triangulate and calculate the XYZ coordinates of hundreds of points along the line.

Using optical CMMs is another approach to noncontact inspection of body assemblies (see story on page 24). These CMMs offer not only extremely fast data gathering but also portability.

Putting noncontact sensors to work

Although noncontact sensors are used in a variety of measurement and inspection applications, applying their technology to the 3-D dimensional control of car bodies on the production line has been challenging. The first relatively successful application of the technology was with a stationary multisensor inspection cell that uses a number of CCD camera/laser light sensors to gather dimensional data. The system meets production cycle-time requirements by simultaneously measuring a number of body elements, determined by the number of sensors installed in the cell.

However, the stationary multisensor approach has some limitations. Due to the large number of sensors required in each cell, it's expensive to manufacture, and maintenance is often complex and costly. The system generates relative, rather than absolute, measurement data and requires gold masters that must be certified and carefully stored to support the periodic certification of cell performance. In addition, the system can't be easily reconfigured to meet varying production-line applications.

The solution lies in combining CMM flexibility with noncontact sensor data-gathering technology. CMMs used for body-in-white inspections are called "measuring robots." These are generally designed as horizontal arms, the only structure capable of reliable dynamics, high accuracy and the long measuring strokes needed for full car-body inspection. The design is also well-suited for adding a second arm to create a multi-axis inspection cell. With such a configuration, cycle time is dramatically reduced, throughput is maximized and measuring errors are minimized.

Measuring robots are fast, accurate and rugged machines designed to operate in the most severe production environments. They can be easily integrated, in-line or side-by-side, with production equipment. They also resist ambient shop-floor conditions such as temperature gradients, airborne contaminants and vibrations. Internal positive air-pressure systems keep critical components clean. Machines are constructed from materials that ensure an optimal rigidity-to-weight ratio and react well to thermal variations.

The addition of an optical sensor to a measuring robot is a solution that combines high-speed data gathering and all the typical capabilities of a CMM, such as easy programming, quick reconfiguration, high accuracy and full consistency of results, with off-line CMMs equipped with traditional probes. It meets production cycle-time requirements by using the noncontact sensor to measure elements based on single-image acquisition through single positioning, and by utilizing the CMM's high-speed characteristics for car-body inspections in production environments.

In a typical measuring routine--for example, the position, diameter and orientation of a through hole in a car body--the system executes the following steps:

Using the CMM's flexibility, the sensor is moved to a position over the hole.

The sensor acquires an image of the hole, projecting it over the part surface.

The 3-D position of the hole is calculated from the deformation induced on the sensor light by the presence of the hole.

The system reconstructs both the hole plane and the hole shape without repositioning the sensor or moving the CMM.

Reaching critical features

Probe orientation is critical for accurate, efficient subassembly and body-in-white measurements. However, part accessibility limitations often require frequent orientation of the head during the scanning process, and unless the sensor is mounted in an articulated holder, results in less effective overall system throughput.

Continuous two- and three-axis servo wrists solve this problem. These are designed to quickly orient the probe to any attitude, following precise 3-D trajectories. The system controller continuously monitors the servo wrists' speed and motion to reach maximum machine efficiency. Servo wrists guarantee full-part accessibility--by extension bars, if necessary, for inside car-body measurements--with no loss of accuracy. The sensor is functionally connected to the CMM through the three-axis wrist and is recognized by the CMM like any other standard probe; therefore, all sensor functions are integrated and managed by the software. Probe changers and magazines are fully supported. The sensor 3-D measurements are referred to the CMM coordinate system. Advanced software compensation routines ensure high probe-positioning accuracy without the need for calibration.

Advanced measurement and inspection software has evolved along with high-speed scanning CMMs, optical sensors and probe holders. Software, such as Brown & Sharpe's PC-DMIS measurement and inspection software, supports both laser and CCD optical sensors. The CMM is controlled by a distributed processing architecture that includes the controller and main computer.

PC-DMIS software manages all machine functions, the operator interface, part programming and 3-D data analysis. It also provides an interface between the sensor and computer for efficient data exchange. PC-DMIS includes special data-analysis tools for automobile body features, as well as special programming that relates data from both arms of a dual-arm measuring robot to a common axis. It calculates the most suitable wrist angle for measurement, based on the theoretical location and vector of a part feature. These features are accessible through a graphical user interface that requires the operator simply to point and click on the program and operation of choice.

Brown & Sharpe is currently working with industry groups to develop an optical sensor interface standard so that any type of optical sensor will work effectively with any type of CMM.

The ultimate value of metrology is its ability to forge a link between design intent and manufacturing capability. CMMs, particularly those capable of scanning, are an extremely flexible and cost-effective way for manufacturers to take advantage of the benefits of metrology for improved quality and process control.

About the author

David H. Genest is director of marketing and communications for Brown & Sharpe in North Kingstown, Rhode Island. Genest has bachelor's and master's degrees in mechanical engineering and has been involved in product engineering and development during his career. His background in metrology system design and development includes the introduction of Brown & Sharpe inspection systems for shop floor applications. He is a member of the Metrology Automation Association board of directors. Letters to the editor about this article can be sent to letters@qualitydigest.com