BLOG
BLOG
Recently, Drishti’s CTO & Co-founder Dr Krishnendu Chaudhury, one of the world’s leading computer vision and deep learning experts with dozens of patents to his credit, was invited by Bangalore Chamber of Industry and Commerce (BCIC) to present on “Applying AI and Computer Vision to Manufacturing Operations.”
In his address, Krishnendu talked about practical applications of the latest developments in computer vision technology and how they can translate into tangible results for manufacturing firms. These developments, in his experience, have enabled manufacturing processes to be carried out quicker and more efficiently. Using real world examples from Drishti’s archives, he demonstrated how these advances have enabled better decision making and led to impressive productivity gains.
Here’s a brief synopsis of the session; the full recording is available below.
How AI and deep learning revolutionize the decision-making process in factories
Making intelligent decisions that work at scale is one of the key areas where machine learning and AI can be implemented. Fast, accurate decision-making is critical in manufacturing, which is tough given that situations are complex and often based on very little data. Companies need intelligent systems capable of simultaneously creating new data while identifying patterns and providing insights 24/7 at 100% efficiency with little room for human error.
Considering how incredibly data-intensive today’s factories are, it is impossible to even imagine crunching through all that data manually. Human beings are just too slow at processing such high volumes of information and may also miss vital parts of the picture during their analysis. This is where AI and machine learning come into the picture.
AI from Drishti uses a type of advanced statistics to find patterns and generate hypotheses that help you make better decisions. And it does this by moving away from what once used to be pure manual activities such as data gathering and decision making. Drishti’s AI models are trained to detect patterns and anomalies by creating and analyzing large amounts of information from video streams, then use that data to help humans make smart decisions, faster than a human ever could.
Five ways manufacturers can integrate AI and machine learning
1. Predictive maintenance
Predictive maintenance can save time and money by allowing equipment to be repaired before it breaks down. Factory equipment could detect within itself that it is due for some kind of repair or update before it actually happens, alerting the maintenance staff to address it before it becomes an issue, and plan the repair or upgrade when it’s convenient. This would allow for less down time and more productivity from factory machines at the same time saving money for businesses.
2. Demand prediction and intelligent supply chain
This is not specific to manufacturing; demand prediction can be used in retailing and other industries, too. The rough idea is to use inputs such as current prices, sales history etc. to predict upcoming sales volume. If we can predict the demand with a decent degree of accuracy, that will tell us how much raw material should be purchased or produced so as to meet the demand without having excess inventory or overproduction. Once you upload data and train your program, the AI can help to make your decisions without human intervention — which means your demand forecasting is ready when you need it.
3. Warehouse automation
Warehouse automation doesn’t simply mean robots carrying boxes around. These are technologies capable of demonstrating human-like cognitive functions to assist warehouse operations while reducing the human touch needed to stock shelves. These smart systems proactively stock themselves by placing orders, talking to delivery trucks etc., even in situations where there are no humans present at all. This provides benefits like predictive stock replenishment, quicker inventory counts, better peak picking processes and increased productivity.
4. Automated defect detection systems
Defects in products are a common problem across industries. Unfortunately, defects can’t be prevented 100 percent of the time, and can cause serious harm to people and property or may reduce the service life and reliability of the product. Using computer vision to detect such defects therefore becomes vital. The introduction of artificial neural networks into industry has boosted the development of intelligent systems that allows you to detect underlying causes, from the design stage right up to delivery of your final product. Engineers and manufacturers can use these systems to ensure that their products are safe and reliable and prevent expensive mistakes while saving time and money.
5. Focus on people
Despite all of the innovation we’re seeing in terms of automation in factories today, a lot of factory work is still centered around people. There is still a lot more flexibility and reasoning in humans than machines. While humans bring these strengths with them, they also bring weaknesses, which not only causes errors, but also productivity issues. Humans can tire out, they're good on some days, they work very slowly on other days. Levels of training among people may vary. Some workers naturally work more efficiently than others. To measure this difference in productivity is a serious challenge many manufacturers face today.
In a factory, a small delay/mistake on one station can cause a significant difference in overall production throughput. So, if we have an intelligent system measuring human actions, that takes notice when a cycle is longer than takt time. This is exactly what Drishti offers its customers. Drishti is actively bringing AI and computer vision — and therefore, many of the benefits you read about here — to manufacturers for the first time.
It’s impossible to condense the entire discussion into one blog post, so make sure you check out the full recording of the presentation.
