Artificial intelligence (AI) is proving to be a game changer in ensuring the quality of automotive components which are complex, price sensitive, high volume and frequently safety-critical. By pairing AI with Machine Vision (MV) it has become possible to inspect every part coming off the line – something that was neither economic nor practical using human operators. This enables a camera feed to be reviewed in real-time and have faulty widgets identified and tagged either physically or virtually. In this article, I will look at two pioneering installations, and point to the lessons that they offer in terms of realising these benefits in the context of a safe and secure connected IT environment.
Audi’s 100% inspection of welds
One of the pioneers of this approach was the Audi A3 line at its Neckarsulm plant. This site has 2,500 autonomous robots on its production line. Each robot is equipped with a tool of some kind, from glue guns to screwdrivers, and performs a specific task required to assemble an Audi automobile. Audi assembles up to approximately 1,000 vehicles every day at the Neckarsulm factory, and there are 5,000 welds in each car. To ensure the quality of its welds, Audi performs manual quality-control inspections. It is impossible to manually inspect 1,000 cars every day, however, so Audi uses the industry’s standard sampling method, pulling one car off the line each day and using ultrasound probes to test the welding spots and record the quality of every spot. Sampling is costly, labor-intensive and error prone. So the objective was to inspect 5,000 welds per car inline, and infer the results of each weld within microseconds.
A machine-learning algorithm was created and trained for accuracy by comparing the predictions it generated to actual inspection data that Audi provided. The machine learning model used data generated by the welding controllers, which showed electric voltage and current curves during the welding operation. The data also included other parameters such as configuration of the welds, the types of metal, and the health of the electrodes. These models were then deployed at two levels, firstly at the line itself and also the cell level. The result was that the systems were able to predict poor welds before they were performed. This has substantially raised the bar in terms of quality.
Digital transformation at Bosch
Another major name in the automotive industry, Bosch, is trialling a similar approach. In partnership with Lynx Software Technologies, Bosch VHIT, the vacuum & oil pumps manufacturing subsidiary of Bosch, is testing a new proof of concept camera-based quality program for use with real-time decision making in industrial settings. The move is part of the company’s digital transformation of its processes and product development. The program captures data from cameras on manufacturing plant floors and logistics warehouses and harnesses machine learning algorithms to identify quality issues and feed information into the MES system, in order to generate an optimal decision in real time. When securely connected to the cloud, the system benefits from continued access to advanced artificial intelligence algorithms and data analytics packages.
Since these systems are critical to the manufacturing process, they need to be protected against hacking and the malfunction of another program running on the same hardware. By partnering with Lynx, Bosch VHIT was able to close the digital feedback loop that is reliant on capturing quality images and analyzing the data to provide a safe real-time action. The LYNX MOSA.ic for Industrial product enables the program to run multiple functions on a single SoC without compromising performance, security or safety.
“As we continue advancing cutting-edge technology applications for factory automation, we are excited to partner with Lynx to accelerate a new, secure IIoT-based quality system for the market,” said Riccardo Sesini, Digital Transformation Manager, Bosch VHIT. “In increasingly connected manufacturing environments, manufacturers require safe, versatile, and resource-conserving solutions. Lynx has a long history of robust, safety-critical, high-availability systems and was the obvious choice to help us realize this latest program in a safe and scalable way.”
Securing the new infrastructure
Central to the success of both installations is the collection and processing of data relating to a mission critical process at the edge (i.e.: on the production line) rather than in the cloud, so that adjustments to the process can be made in real time. For these new manufacturing quality systems, the LYNX MOSA.ic for Industrial framework is focused on ensuring security and mitigating any period of equipment downtime that could impact business output. It provides a software platform that can run the inference engine and control functionality on the same platform, ensuring that these applications are appropriately isolated and allocated the right hardware permissions (and nothing more) to perform their tasks. A camera might highlight an issue; then a soft PLC can then connect to the line and make appropriate process improvements. We call this infrastructure the mission-critical edge.
The Lynx framework consolidates mixed criticality workloads running on the same multicore processor – the resources and performance provided by the hardware platform, and the capability of the software components. At the same time, it completely isolates critical applications from non-critical workloads, in order to provide high levels of immunity to the former from cyber-attacks. Additionally, this greatly reduces the architectural complexity, cost, and number of points of failure – a critical factor in ensuring business resiliency.
Many manufacturers are exploring better ways to proactively and continually improve quality. Whilst these approaches may generate some false positives, this is much better than failing parts making their way to end customers. As the algorithms improve, the incidence level of those occurrences will reduce. The combination of AI and MV is significantly more effective than batch testing, which is used to manually and retrospectively trace faults back to the manufacturing environment and workers to understand root causes and make changes to processes. However it is important to appreciate that this approach relies on the availability of robust edge solutions for the connected camera-based quality system that enable real-time responses to be delivered to events while ensuring critical applications run reliably and safely alongside other functions operating on the server / gateway hardware.
Pavan Singh is Vice President of Product Management at Lynx Software Technologies