top of page
  • Writer's pictureBoaz Ein-Gil

AI for Manufacturing

Updated: Oct 17, 2021

Large variety of shoes to choose from
Limitless variety of shoes

Current Trends

In recent years, consumers have been gradually accommodated to having virtually unlimited variety of products to choose from, allowing them to be picky with their purchases and buy products tailored to their needs and wishes. The key factors that enabled this change are global supply chains and e-commerce – no longer are consumers constraint by the products offered in local brick and mortar stores - no longer do they need to compromise. In addition, the proliferation of customization options for products (ranging from cell phone cases to cars), makes the number of varieties of each product line very large.

While this trend makes consumers happy and generally increase demand for products, it greatly impacts retailers and the supply chain. Moreover, manufacturers are faced with the challenge of producing large varieties of products in relatively small batches (high-mix, low-volume).

Automation to the Rescue…

Production line automation is key in getting the necessary efficiency in modern factories. It helps push yield higher and increase throughput while reducing the dependency on factory floor workers. However, automation does not come cheap. There are high upfront costs for hardware and integration services in order to build or convert an existing production line to use automation, even when you are aiming for partial automation. This cost can be justified when the line you are building will be used for a long period of time to produce large batches. Product variety might not have a negative impact on the cost-effectiveness of automation in cases which variety does not translate to changes in the production process (e.g. producing multiple colors of the same product). However, when the variety does require changes (e.g. variety of sizes for the same product), re-implementation of automation might follow, leading to additional costs

Leveraging AI

So how can AI help? AI solutions provide a level of reasoning that is essential if you like to cope with changing conditions. Adding AI to the automation would make it easier to adopt to changes in the line and greatly lower the cost of changes. Here are a couple of examples to demonstrate the benefits of this approach:

AI for Vision

While computer vision has been part of manufacturing automation for a while, moving to AI based vision capabilities (‘visual cortex’) allows for quicker adjustments that can be done on the assembly line with relatively quick turnaround time. Instead of building specific computer vision solution, you can leverage a pre-trained AI agent and perform specific tuning at the last hop with a small number of examples getting you up and running in a matter of hours rather than days and at a reduced cost. When it comes time to change the line, re-training can be performed quickly.

AI for Planning and Controlling

Currently, building automation for a production line require building specific automation code to operate the robotic elements (arms, etc.) in order to perform a specific task. In some cases, you could leverage visual software tools to produce this control code (i.e. visual programming), but at the end of the day, the robotic elements follow a very specific set of instructions.

When you make changes in the line, you need to reprogram the robot in order to adjust its behavior to the new environment/requirements. Re-programming is costly in term of time and in many cases, it requires bringing in outside consulting which is an additional expense.

Alternatively, leveraging AI Agents technology, you can train an Agent to perform the task required in a simulated environment (controlling the movement of a virtual robot in a virtual station). During this training, the AI Agent learns how to perform the task well and this knowledge then can be used to control the actual robotic elements on the actual line. When line changes are required, you can adjust the simulation accordingly and retrain the Agent. Furthermore, in some cases, retraining might not be required at all since the Agent learns to perform the task in a less prescriptive way that allows small adjustments and increases robustness.

What’s Next?

Modern consumer demand for variety of products can fuel the growth of factories, but only for those who keep up. Leveraging AI is essential for a manufacturer looking to become more flexible and dynamic and ride this wave. You can start the journey of transforming your business with small steps today, and we, at Agent Factory, will be happy to assist – drop us a note and let’s start the conversation.

16 views0 comments

Recent Posts

See All


Commenting has been turned off.
bottom of page