Big news: Tryolabs joins Qubika. Same teams, broader set of combined capabilities. Read now ->

Case study

Automation of manufacturing processes

27% operator throughput increase, saving 10 seconds per foot scan processed, while 2/3 times our system matched or surpassed technicians' results.

Industry: Manufacturing
Client
Solution: Computer vision

Process automation improvement

Foot Levelers is the world-leading provider of individually designed functional orthotics. They have improved millions of patients' functionality and quality of life. We engaged with them in their search for improving the process of building custom orthotics.

Challenge

Industry 4.0 players are always in the search of improving their operation's efficiencies. Foot Levelers wanted to explore manual process automation by incorporating Machine Learning solutions and speeding up the manufacturing process.

Every custom design process involves a set of particular tasks for each client. Even in this scenario, some tasks are quite repetitive and others have the opportunity to be simplified or done more consistently. Processing over 1000 orders per day, Foot Levelers had a vast amount of data available that could be used to assist in the insole design process. Along with the patient's data, the company also has scans of the soles of the customers feet taken with their own system, and historical information from previous orders. All this could be used to understand what patterns drive Foot Levelers technicians to make their decisions and assist them throughout the design process.

Solution

We helped Foot Levelers to automate their manufacturing processes by applying Machine Learning and Computer Vision solutions. Measurements and placement algorithms allowed us to enhance the custom orthotics design.

We identified an opportunity to automate the work done on top of the foot scans with Machine Learning and classical computer vision methods. Previously, technicians had to measure the feet by hand and place pads manually.

This solution had to match technicians' performance and would let us evaluate its impact on the manufacturing process.

27% increased

throughput

Discover our approach

The system is based on multiple keypoint detection algorithms that are able to automatically locate the relevant spots within over thousands of foot scans processed every day.

We developed a system to automatically detect 4 key points from a foot scan: top ("highest" toe), bottom (heel), left and right (with respect to metatarsal heads). These key points can then be used to measure the length and width of the foot, information necessary to select the insole's size.

Metatarsal and medial arch pads are corrections present in every insole process, and they must be placed on a specific part of the insole to be effective. We leveraged our key point detection system to automatically position these pads on the insole by looking at the foot sole scan.

These systems were designed for on-premises deployment, running on Foot Levelers own infrastructure with minimal computing requirements. This solution was also integrated into their insole design software, making it easily accessible to all technicians.

Tryolabs made implementing AI so simple. Our updated software has become so much faster and easier to use. Our operators have increased throughput by 27% so far.

Chad Warren, Director of Engineering | Foot Levelers Inc.

Discover how the world’s leading enterprises partner with Tryolabs for building data-driven solutions that boost their KPIs.