BLOG
BLOG
2021/06/29 | Time to read: 3 min
エリン·コールドウェルは、十数社のSaaS企業の市場参入戦略を立案·実行してきました。彼女は、説得力のあるストーリーテリングと顧客中心主義の組み合わせで、ドリシュティの価値を市場に示す手助けをしています。戦略的マーケティングの修士号を持ち、ボストンに在住しています。
Last month, Drishti’s Founder and CEO Prasad Akella joined numerous other experts in the world of AI for EmTech Next, MIT Technology Review’s annual conference that this year explored leadership strategies and the new technologies that will power a forever-changed workforce.
In his session, Prasad explored the world of computer vision and AI, going beyond object detection and fault inspection to analyzing video streams and understanding humans at work.
Here’s a quick overview. The full session is linked below.
Let’s start with two myths that are pervasive in manufacturing:
First, the myth that robots make all of the stuff that we buy.
From cars to smartphones. The shocking fact is that it is people who make stuff happen in the world. And what is more wild is that we have no idea what these people are doing. This becomes very clear when we have massive product recalls from cellphones to airbags.
Second, the myth that the primary path to productivity is automation.
There’s an idea that people are optimized, and the only way to get more efficiency gains is to swap a human for a robot.
Let’s take manufacturing, for example. Here’s the math: There are 345M people around the world on production lines. We have 2.7M robots currently — about 300k in the U.S. One robot displaces ~6 people, and there are +373K new robots deployed annually. Each robot displaces 4.5 people, with a ~0.5% annual displacement. That means at this rate, it will be 153 years before there are more robots than people in manufacturing.
Humans aren’t going anywhere. So how do we optimize processes centered around people?
There are two steps on this journey to empower people:
The first is to solve the measurement problem by automatically getting data from manual assembly lines. As we all know, you can’t improve what you can’t measure.
Drishti has developed a new technology, action recognition, that parses video using AI and computer vision. Like Siri parses the spoken word, we extract process information from the video stream. As you can imagine, video is orders of magnitude more difficult to analyze and draw insights from than a single frame.
By tying the data to the raw video, suddenly we have deep insights into what happens on the production floor.
The second step is applying this data to solving real world problems.
We applied this technology to the stodgiest of industries — manufacturing, which is 15% of GDP and sorely in need of disruption. Just think about the fact that some 113 years since Henry Ford and the Model T, we are still gathering data on manual lines using humans and stop watches.
Here’s a real world example of the power of these two steps in action: Action recognition technology identifies cycles and even individual actions on the assembly line, at every single station, creating the baseline data.
We process this massive volume of data and present the raw facts, as well as the synthesized insights, to the user in a way that is actionable. With the focus being on acting on the small volume of process issues. We help manufacturers target their scarce production, quality and Industrial engineering resources on what matters.
And the results are clear: customers are seeing as much as 11% in efficiency increases, 30% reductions in defect rates, 15% reductions in scrap rates, kaizen events taking half the time and saving $10K per defect.
These are industry-changing results that put people back in the driver’s seat, letting higher wage countries compete on efficiency and technology — technology that works with people versus replacing them.
Metrics and measurement aren’t new, but fundamental change is possible with new technologies using the new data set that we have just created — automatically, at a massive volume, without bias.
And these developments have broad implications beyond manufacturing. The opportunity is vast — essentially, applicable wherever people are at work, including warehousing and distribution, food service, healthcare and pharmaceutical, retail and more. People will continue to do the work across industries, but with technology like AI and computer vision augmenting them to everyone’s benefit.
