In this project, dMACQ Software has used robotic process automation (RPA) with artificial intelligence and machine learning for automated testing of industrial machinery components, for an automobile industry customer. As part of the project, an Android mobile application has been designed as a front-end. The technology in the back-end uses AI based computer vision and machine learning to recognise machine components.
At the outset, the testing process at the client site has always been done manually and involved creation of several paper based information. The challenge faced by the client was to eliminate the paper trail, the human element in testing and fetch immediate results and cut down the test processing time.
Use of computer vision technology recognises image patterns on a machinery component, accurately recognises the contours and calculates the various dimensions of the contours which are relevant to perform the test. This is otherwise performed by trained human test operators. Another innovative use is the comparison algorithm. dMACQ uses advanced comparison algorithms to calculate deviations using multiple parameters to accurately determine the fault percentage compared to the allowed standard deviation. The use of advanced calculation eliminates the dependency on humans.
dMACQ provided an innovative mobile app based solution which connected to SCADA system with a simple bar-code based integration. By connecting the mobile app to the SCADA system enabled identification of the component being tested and the result being relayed back instantaneous was also achieved using bar-coding technology. This ensured zero integration cost to the client or any need for any additional technology.
Post-deployment, as soon as the component comes to the test station, the test operator uses the mobile app to scan the barcode affixed to the component. This identifies the component and also sends a signal back to the SCADA system. The operator uses the mobile app to take pictures of the components. The mobile application comes with standard in-built guides which ensure that the picture for each of the component cannot be taken randomly and has to meet strict guides.
This eliminates human error in taking pictures using mobile app. As soon as the process of taking pictures is completed, the mobile immediately renders all the test results on the mobile app in seconds and generates a bar code with the test results. The operator flashes the bar code to the bar code reader of the SCADA system and the test result is directly fed in to the production ERP. The component is now moved to its next logical step. The entire process is completed in about five minutes of arrival of the component in the test station.
The company carries out about 2250 tests in a day. Each of the test uses about 25 pages. There is straight saving of about 56,250 pages per day. There have also been significant time savings; reduced from an average of 60-75 minutes of combined processing time to less than five minutes. This has increased the production capacity and allowed redeployment of skilled resources to more productive use.
No third party vendors vendor or products were used. The entire application is built inhouse by dMACQ.