QMT Features: November 2013
Inline robots for QC
A Swedish Government funded project, InRob,  with Volvo Car Corporation has successfully demonstrated  in-line robot inspection for automotive Body In White. By Roger Holden, Nikon Metrology.

InRob was a Vinnova (Swedish Government) funded project with Volvo Car Corporation. Other partners included, Scania, LK Scandinavia, Fraunhoffer Chalmers Gothenburg,  Chalmers and Linkoping University, and was supported by Nikon Metrology UK.

  The project outline was for future production control in Body In White,  delivering proof-of-concept that an in-line robotic inspection system could be deployed in a fully automated / off-line programmed way, and would meet the needs of accuracy and repeatability in Body In White assembly. The platform used was Nikon Metrology’s robot integrated laser scanner (MMDx100).

Starting with CAD nominal design (in this case CATIA V5) Nikon Metrology supplied Chalmers with details of its proprietary feature definition format. This gave a very fast, open and simple transfer from Volvo’s own metrology planning toolkit (GD&T) into Nikon’s analysis software (Focus 10).

Working closely with Linkoping University and LK Scandinavia, Nikon developed best practice guides including optimal approach and angles for laser scanning. This was used by Chalmers to optimise the measurement sequence and give base robot program for final verification in either a robot vender specific simulation system or generic 3D CAD based simulation system (using Realistic Robot Simulation technology). In this project both ABB’s Robot Studio and Dassault’s Delmia V5 were successfully used, providing a simple method of creating robot programs and proving them off-line.

Scanner technology

For on-line inspection Nikon Metrology’s MMDx100 laser scanner was used, including its Enhanced Sensor Performance (ESP) technology, changing the laser power real-time for consistent data acquisition on shiny and matt surfaces without data loss.

Working within the K-Series optical CMM working volume, the scanner acquires data where needed, in any robot orientation, in one unified coordinate system. This gives great accuracy with low data sets, meaning fast processing. The point cloud data is acquired within the robot’s reach inside the large volume of K-series optical CMM.
For ultra-high speed inspection the acquisition can be taken on one PC and the analysis on another. In this case, the analysis can take place at the same time as the next scan. This also facilitates multi-robot inspection in a single station; with each robot doing its own acquisition but then passing the data to station PC to do multi-robot inspection analysis.

The analysis is run as an automated script, which is programmed once off-line, as a simple sequence within Nikon’s Focus software. The nominal features are then compared to the filtered point cloud. This starts as an alignment based off scanned tooling balls on the jig. This puts all the measured features in the same coordinate system as the car. Then the rest of the features are auto-detected to CAD nominal.
Integrated Solution

Putting this all together we successfully demonstrated the automated workflow with the Nikon Metrology Automation Software at the hub of the robot cell
There is an automated route from CATIA, through the metrology planning toolkit (GD&T) into Nikon’s proprietary feature definition format (.mff) over local network. Rules for scanner orientation angle, speeds, directions are preset-based on best practice. These are used in Chalmers software to optimize measurement sequence, outputting to standard robot simulation package to complete all programs off-line – again transferring full syntax programs over local network to robot cell.

On starting a robot program the Nikon Automation Software manages the interface between robot and metrology system, based on its “Adaptive Robot Control” (ARC) technology. The acquired point cloud is then automatically processed – filtering, auto defining alignment features, aligning, and automatically comparing point cloud to nominal features. Features inspected were a range of 2D holes/slots; surface intersect points, and 3D weld bolts, Christmas trees and T-Studs. Locally the features are checked against tolerance warning limits – for local cell alerts, and output over local network in PCDMIS format for bodyshop Statistical Process Control system

The repeatability of measurements was 60µm 3ó; with spatial accuracy within 100µm within the robots working volume. Final inspection accuracy compared to CMM touch probing was 150µm for 2D features such as circles and slots; and 400µm for weld bolts and Christmas trees – again all 3ó. This was all achieved in single pass at high speed. Of course more passes and slower speeds improve accuracy – but the trade-off is longer cycle times. In this deployment inspection times were approximately 2 seconds/feature.

The solution is available today on any robot platform. Using Nikon’s ARC technology it’s possible to port to any robot in less than two weeks. Currently there are fully working interfaces with Fanuc, ABB, KUKA, and Siemens PLC. Nikon continues to develop its fitting algorithms which are shared across all technology platforms.l
You can now view all QMT Magazine issues on your favourite tablet or smart phone.
Download the free Quality Manufacturing Today App from the Apple iTunes App Store or from QMT Magazine on Google Play.

Rob Tremain Photographer
Click above to see full page display and links to QMT articles.
View Olympus video>>
Vision 2016 ad
Bowers logo
Mitutoyo logo