5S tool board systems empowered with machine vision AI neural network to control tool availability. Such that no item can be checked-out or missing from its place in a workshop. The solution can be transferred to other industries requiring ongoing object recognition.
1 → 9
Nearly all manufacturing and service companies are known to struggle with the very annoying problem of missing tools.
In good manufacturing practices such as LEAN manufacturing or Six Sigma, this problem is circumvented by using 5S shadowed boards. These 5S boards provide visual management and highlight items/tools that should be replaced/returned after being used. They provide a visual indication when a tool is missing (e.g. at the end of a shift).
However, in workshops that provide access to multiple users, 5S boards typically fail to keep tools in their place.
Machine vision service development
Backend software development
Spring Boot framework
Netty client-server framework
Docker software platform
Gradle software build tool
EnCata offers comprehensive hardware and IoT product development services at a fraction of the cost
• R&D + design + manufacturing under ONE roof • Scale up and down your team • Intergrated hardware + software development • New technologies and research
To solve this problem, EnCata’s AI software team suggested the use of machine vision (MV) in combination with the standard 5S tools boards. Using a CMOS camera connected to the cloud server with AI/ML helps to monitor if tools are in their place.
EnCata’s developers created an artificial neural network (NN), which was able to capture and analyze the live image from the camera hanging above the 5S board. The NN went through a supervised learning routine of image recognition, taking into account camera lens distortions.
An item taken from the 5S board has to be checked-in and out by scanning a QR code of the tool. As a result, any workshop user (with a user account in the ERP/CRM system) automatically registers the tool when taking it for use.
In the case when any tool is taken from the board without checking it out, an ‘Andon’ signal makes a sound+visual alarm. The neural network recognizes all the objects/tools on the 5s board and understands which tool is missing, enabling dual (visual + AI) control of the item’s displacement.
Such AI/MV software can be implemented in various industries and production facilities embracing the LEAN approach. For example, inventory Kanban stock with an automated alert for component procurement is the next “low hanging automation fruit” that can suit the automotive and aerospace industries.
A very powerful solution for manufacturing workshops was developed in this case. The AI/ML software significantly reduces the time spent searching for items. The backend system ‘talks’ to the ERP inventory management system and helps to gather analytics for tools utilization by each user.
The solution enables a plethora of other applications involving object recognition.