The New Super Power for Supply Chain: Computer Vision
“Some people call this artificial intelligence, but the reality is this technology will enhance us. So instead of artificial intelligence, I think we’ll augment our intelligence.” — Ginni Rometty, served as chairman, president and CEO of IBM
The new era of visibility in logistics
In 2010 IBM was already talking about “Cognitive Supply Chain”, an operations model where each node in the supply chain is instrumented, interconnected and intelligent. Its relevance is such that in the next three years it is estimated that 46% of supply chain and logistics executives will invest in AI/cognitive computing as a priority.
The application of Artificial Intelligence and Machine Learning is commonly used in virtually all industries, companies like Facebook, Google and Netflix use it to recommend content to their users; Tesla in the driving of autonomous vehicles; Amazon in the management of its warehouses and financial institutions such as Klarna, Coinbase and others for identity validation processes and authentication processes.
In 1936, Alan Turing laid the foundations of computer science, which in turn cemented the origin of the first cases of artificial intelligence. He first described in his paper “On computable numbers, with an application to the Entscheidungsproblem” the concept of algorithm — the ordered set of systematic operations that allows to make a calculation and find the solution of a type of problem. In 1941, Konrad Zuse created the first fully automatic, programmable computer, the Z3. It is considered the first computer in modern history. However, it was not until 1956 that the term Artificial Intelligence was first coined at the Darmouth conference by computer scientist John McCarthy.
A special case that leverages AI, ML and elevates it to Deep Learning is Computer Vision. Artificial Intelligence is a set of algorithms that mimics human reasoning and replicates human learning mechanisms. At a higher level of logical reasoning is Machine Learning, which enables the “machine” to learn through algorithms that are not necessarily explicitly programmed. At a third level is Deep Learning, which recreates the process of a computer brain, weaving neural networks by developing an iterative data architecture.
The three previous stages are the basis of Computer Vision. This is a technology that using algorithms and pattern recognition (commonly matrix processing of binary patterns or through color identification) manages to associate them with shapes, figures and visual structures that can be associated to a human context. Taking this information as the actionable basis for which, it has been programmable, whether in less complex actions such as sending alerts or notifications, to making the decision of what maneuver to perform in an autonomous car.
First the raw data is accumulated and organized. Then each image is categorized, extracting its physical characteristics by means of labels. Next, the software analyzes these labels. The next step is to create a predictive model, and this is how the model finally decodes the images, emulating human vision. In this way, the image processing is converted into information that is used by the “machines” to make decisions and take actions.
The supply chain is one of the main industries that have benefited from Computer Vision. Among its main applications are:
1. Efficient manufacturing
Manufacturing areas are always looking for efficiencies. Computer Vision technology supports elements such as sorting, assembly and quality. For example, Instrumental is a US company that identifies wrong items within the production line. Invisible company enables high-precision assembly, increasing image quality and reducing process time. Lincode enables manufacturing lines by integrating low-cost cameras to detect faults in the production line.
2. Autonomous distribution
The objective is to streamline the supply process in the supply chain, either having a high fulfillment in warehouses or direct to the consumer. The main element of this automation is provided by autonomous vehicles. In the long haul, Locomation are two examples of companies that specialize in long haul and heavy haul. In the middle mile, companies such as Gatik are developing fully autonomous vehicles (N5), which they use to supply the warehouses of Walmart in the US and Loblaw in Canada. But it is in the last mile where there are diverse examples with a differentiated value proposition and multiple applications such as Nuro, Kiwi, GoodieBoxx and Ottonomy.
3. Inventory optimization
The need for shrinkage reduction, better skus control and inventory improvement has led to the implementation of technologies such as Pensa, which uses camera-enabled drones to read and process inventory at a point of sale or in a warehouse. Milkyway, which uses cell phones to audit skus. Vimaan, which uses drones to track events occurring within a warehouse. Unbox Robotics, which has small robots that move inventory racks from one place to another in the warehouse.
4.Risk prevention
Any process within the operations of a company involves risks that may not be visible to the human eye, in addition to not having enough personnel to be present in all areas and processes. So, companies like Stroma use surveillance cameras as a source of information to process images of facility usage as well as personnel risk behaviors. Roomie, although it works directly in warehouses moving inventory, its focus is on reducing the number of personnel moving freight via autonomous forklifts and preventing incidents in the process. The company Kineticeye, which identifies potential unsafe areas within the production plant infrastructure. Cipia, an Israeli company, which through visual telematics alerts drivers when they are falling asleep and measures their driving behavior, helping to assess their performance in a timelier manner and improving driving risk mitigation.
While it is true that the application of Computer Vision is of great impact in any of its applications, companies such as DHL in its logistics radar, consider that autonomous cars are perhaps the application that will take the longest to be adopted on a massive scale, 5 to 10 years. While less complex processes such as robotics may take less than 5 years for mass adoption.
There are multiple ways to optimize processes, Computer Vision is one of them. However, according to McKinsey, to see a tangible result, the most important thing is less related to the technology used and more related to the analysis of the processes where there is the possibility of adding value; the change in the personnel in charge of administration and the objective design of the operating system to be reached.
Over the years, technology has evolved and improved. Today, artificial intelligence combined with Machine Learning and Deep Learning have enabled the advanced development of Computer Vision. Without a doubt, the supply chain has been one of the main areas to benefit from this development, and there are many other industrial sectors that have had a positive impact using this technology. We are entering an era where Computer Vision will become increasingly common in multiple processes used by multiple agents.
Hector Shibata Salazar, adjunct Professor at EGADE Business School and Director of Investments and Portfolio at AC Ventures Fund
Ana Maury Aguilar, VC Investor at AC Ventures
ACV is an international Corporate Venture Capital (CVC) fund investing globally in Startups & VC funds.
Stay updated about Venture Capital, innovation, entrepreneurship and more! Sign up for AC Venture’s monthly Newsletter.
Follow us on LinkedIn: ACV_VC
Follow us on Twitter: acv_vc
Follow us on Spotify: ACV_VC
