Industry : Automotive

Category : Software Development

The Challenge:

To rapidly develop a reliable machine vision system for the automated inspection of brake drum assemblies. The inspection was complicated due to the variations in texture, glossiness, color and shade of the components.

The brake drum assemblies had to be inspected for the following:
• Presence of various components
• Reversal of springs
• Proper locking or engagement of components
• Direction of assembly of geared components
• Angular orientation of clips
• Gauging of diameters, lengths & thickness
• Position of brake linings
• Presence of threading in certain components
• Presence of lettering

Earlier this was being done by using pattern matching which proved to be inadequate to handle the variation and complexities involved in inspecting the assemblies.

Technology:

Hardware : Sony IEEE 1394 FireWire Camera, PXI/Compact PCI, Polarizers, High frequency lighting
Software  : LabVIEW, NI Vision development module

The Solution:

The brake inspection system designed by Soliton was fully automated using the latest in Virtual Instrumentation and Machine Vision technology. Custom built algorithms that could self-adapt ensured reliable inspection irrespective of the wide variations in the components. The four quadrants of the brake assembly were imaged separately so as to increase the resolution. These images were then analyzed by the custom software using various techniques and the marking mechanism was then actuated based on whether the drum was accepted or not.

Engineering Challenges:

• Scalability - The system had to be scalable to test various models of brake drums, which was one of the most important considerations of the customer.
• Positioning of components - Detection of components was also affected by the presence, absence or shift in position of the components in the background which posed a major challenge while designing adaptive processing algorithms.

Value Addition:

• Configurability - The user was offered complete flexibility in choosing the features to be inspected along with the inspection criteria and also changing from one model to another.
• Image quality - Software controlled motorized zoom lens, Polarizers and High Frequency lighting were used effectively to ensure that the images taken were most suitable for an error-free analysis. Fig 1. Shows the brake assembly mounted on a fixture to acquire the images.
• Reliability - Every algorithm that was created or modified was validated against hundreds of brake assembly images to ensure the reliability of the algorithm over the entire sample set.
• Statistical analysis - Statistical details accumulated on the various defects provided the customer with valuable information regarding the problems in raw materials and upstream processes, which could then be curbed at the source itself helping to improve performance significantly at the customer's end.
• Customized reports - The reports could be easily customized and accessed over the network.

Commendation:
This application was adjudged 'The Winner' under the automotive category for the NI Week 2003 Best Application Contest.
http://sine.ni.com/cs/app/doc/p/id/cs-367