Artificial Intelligence (AI) has already exceeded human performance on visual tasks. Salman Chaudhary, CEO at EmpiricAI, looks at why that opens up huge potential for manufacturing businesses.
“Industry 4.0 will see smart machines ushering in a new era of manufacturing opportunity with the fusion of physical assets and AI having a positive impact on every element of a business. Manufacturers are already changing the way products are produced, moving towards automated processes, with more adopting advanced technologies, such as AI. One of the areas with most potential for leveraging AI automation solutions is Computer Vision (CV).
CV is a field of artificial intelligence that trains computers to interpret and understand the visual world. Using digital images captured from CCTV and applying the technique of ‘learning’, the platform can more accurately identify and classify objects or data and then react to what it sees.
Here are the top five uses of Computer Vision Artificial Intelligence in the manufacturing sector:
1. Defect detection
Manufacturers have always struggled to achieve 100% accuracy in defect detection and that adds unnecessary costs. The simple fact is that humans don’t perform well on monotonous task such as inspecting products especially over an eight-hour shift.
Visual inspection is therefore visually and cognitively challenging for people, especially where there is a rapid throughput of product. Where a person is inspecting hundreds or even thousands of products for scratches or other imperfections on a high-speed production line, the repetitive workload will quickly lead to degradation in judgement and, ultimately, in faulty products being delivered to customers.
Conversely, a Computer Vision AI solution, such as WorkSafe Analytics, gathers real-time data from CCTV and analyses the data streams based on the pre-set quality standards. Any defects are instantly identified and an alert or notification is generated. Defective items can be dealt with in a number of ways including automated removal.
CV AI platforms also keep a log of defects, allowing management to identify trends over time, which can be linked to specific shift patterns, raw material suppliers or environmental conditions. Being able to take corrective action in these areas means that CV AI can deliver a rapid return on investment. The real potential is in precision manufacturing and advanced engineering applications where the objective is 100% quality control.
2. Stock taking
Computer vision AI can be used in stock counting, keeping an accurate record of product by individual Stock Keeping Unit (SKU). The process is carried out autonomously using CCTV. When a SKU drops below a pre-defined inventory level in the warehouse, an email alert is sent to the management team in order to replenish.
3. Improving Health & Safety
Manufacturing remains one of the most dangerous sectors to work in. According to the HSE (Health & Safety Executive), an average of 85,000 workers suffer from work related ill health each year. Workplace accidents continue to occur despite decades of investment by manufacturers and regulation by the HSE, because a lapse of concentration, an oversight in PPE wearing or fatigue can all lead to an incident.
Manufacturing companies routinely use CCTV and manual inspections to monitor employee adherence to their H&S policies. This approach does have a major weakness, though, because it relies on a person constantly monitor the video stream. This is prone to human error caused by tiredness or boredom, with the consequences, being that a worker who should be wearing PPE in a hazardous area being missed, for example.
CV AI overcomes this by constantly monitoring the entire workspace for correct use of PPE. If there is a violation; a person not wearing a safety harness when working at height, or failing to wear safety gloves, goggle or masks at their workstation an alert is sent and intervention can be made. Moreover, HSE managers can gain insights on non-compliance trends and patterns that allows them to conducted specific education and training and take targeted actions.
4. Packing accuracy
Ensuring the accuracy of multiple products in a single outer is an area where CV AI can add real value. A typical example would be the design for a packaging line in a pharmaceutical firm where it is critical that the number of products in the outer is consistent. Any variation obviously has implications for patient health and financial costs for the manufacturer.
The same applied for retail and grocery, where thousands of outers with multiple inserts could have financial implications if a mistake goes undetected. Again, performing the tasks manually can result in error.
In these scenarios, CV AI can be incorporated on an existing production line or run as a standalone unit. The system verifies, at speed, the correct number of items have been packed. Any variations to this are identified with an alarm system. A historical log of the data allows trends to be identified and corrective steps need to be taken.
5. Footfall heatmaps
CV AI can be used to create heatmaps showing the general footfall of employee/visitor traffic, which provides valuable information when looking to improve the spatial layout of a production area. Equally, it can be used in an office environment to show where maximum footfall is occurring. These kinds of insights can be useful for devising HVAC solutions and, with the recent pandemic, for maintaining social distance and reducing the spread of infections.
Computer Vision AI provides design engineers with the ability to create assets that ‘see’ better than humans. The scope of that extends from inspecting manufacturing processes and finished product for non-conformity and defects, through to warehouse inventory.
Whilst IoT is revolutionising the manufacturing sector and automating processes that until relatively recently were seen as in-improvable, it is CV AI that can eliminate the need for human inspection and monitoring. The most exciting aspect of CV AI is that there are capabilities and applications of the technology that remain unexplored. That is a tantalizing prospect for manufacturers that become early adopters of the technology and apply it in ways that help them create a competitive advantage.