An Automated Optical Inspection (AOI) system that improves the inspection productivity of a medical device dedicated to treating chronic lymphedema patients.
Our client started their journey on a mission to make a difference for people diagnosed with lymphedema and venous diseases . Once Koya Medical was founded, they dedicated their time, efforts, and research to creating devices curated to treat chronic venous insufficiency and lymphedema patients. Koya’s goal is to reimagine and transform venous and lymphatic care through innovative, people-centric technologies. This is why they have introduced Dayspring, the innovative multimodal treatment that provides static compression, active compression, and muscle pump contractions — all key components to venous and lymphatic health.
The Dayspring device utilizes a multitude of subassemblies called Flexframes – the actuators responsible for compressing and moving fluid in the body. Each Flexframe is handmade which introduces the opportunity for error during assembly.
To ensure a high standard of quality, each Flexframe requires meticulous human inspection to ensure no defects were introduced into the final assembly. For operators, this inspection is quickly fatiguing, time-consuming, and still vulnerable to human error. To remedy this, the Koya Medical team reached out to Realtime’s expert engineers to create an inspection system to automate the inspection process, which drastically reduced the potential for misses while increasing inspection throughput.
The handmade Flexframe, developed by Koya Medical, can contain up to 50 distinct points of assembly errors or defects. Currently, Koya has to rely on human visual inspection to inspect up to 1,000 assemblies each day. Due to the size of the features in the product being inspected, this task is visually demanding, leading to the potential of misses or latent defects that escape the inspection process.
The Automated Optical Inspection (AOI) system developed by Realtime captures images of the Flexframe using a smart vision camera, inspects all failure (defect) points along the Flexframe, and indicates to the operator if a defect is present. If a defect is detected, an image of the part is displayed indicating where the defect was identified. The operator then has the ability to divert the part into a rework path or to reinspect the part to confirm the defect. This eliminates the human error potential at each inspection opportunity and also reduces the time required to complete the inspection process of a particular batch of parts from minutes to seconds.
Realtime keeps a sharp focus on the end-user experience for all product or automation system development efforts. For this development project, Realtime needed to ensure the user has proper visibility of the workspace while also facilitating the natural flow of parts in and out of the visual inspection process. This included implementation of a go/no go indicator light intentionally placed in the operator’s area of focus to ensure their attention is not diverted while the inspection process proceeds.
With all product or equipment development, Realtime de-risks the design with early prototype and testing; a proof-of-concept in some ways to ensure all user / stakeholder’s need and resulting requirements have been elicited. For this AOI system, the Realtime team set up the camera and lights using a 2×4 frame, other existing materials in the lab, and zip-ties. With little investment up front, the team was able to determine key design inputs as well as demonstrate to Koya the functionality of the inspection station in a geometrically representative system.
After seeing the prototype setup and inspecting sample parts, Koya mentioned, “We didn’t think a vision system would be able to catch these potential defects, but obviously we now see that it can.” This prototype was necessary to ensure client confidence in the system’s capabilities before committing to the large capital expense related to the full inspection system development.
Realtime’s Automated Optical Inspection (AOI) station developed for Koya Medical works as an efficient and reliable inspection system. The key component of the system is a smart vision camera from COGNEX, placed a fixed distance away from the product to capture the full visual field. The product to be inspected is presented to the inspection system by being placed in the visual field. Within 2-4 seconds, the go/no go indicator light is lit either green (go) or red (no go) to indicate the result of the inspection operation to the operator.
During the time the product is being inspected, the 12-megapixel camera takes 3-4 photos per second and the machine vision algorithms compare the actual product visual data to the stored expected visual data. After data alignment between the expected and actual visual data, any discrepancies in the actual visual data are flagged and the inspection result is logged and indicated to the user as a ‘no go’. Lack of discrepancies between the actual and expected visual data is indicated to the user as a ‘go’ or ‘pass’.
The 12-megapixel camera provides sufficient resolution to decipher the fine product features in the field of view. Realtime engineers programmed the vision system to identify the part configuration, analyze up to 50 inspection points in the product geometry, and indicate the process result to the operator. Within the actual product visual data (image), defect areas are highlighted to the operator and displayed on a nearby screen for clear viewing.
In all equipment design, the user interface and user experience is equally as important as addressing the system’s technical challenges. During the development of the Koya AOI, engineers were
consistently evaluating design solutions by role playing the operator. Early concepts of the design included evaluation of various light bar placement locations.
Another aspect of the design that was implemented to consider human factors was the pass/fail light. This was a less obvious user consideration. The light was strategically added and placed in the operator’s normal line of sight to prevent the need to look at the computer for this information. Requiring continual defocus and refocus of an operator’s attention can be fatiguing and make the equipment burdensome to use.
Realtime maintained transparent communication with Koya Medical to ensure challenges and barriers encountered during the design and development of the inspection system were promptly confronted. Obviously, getting a system using cameras, sensors, and software working together to achieve a desired operator workflow is the main challenge.
Additionally, in vision systems, changes in ambient light, as well as light pollution, can cause unwanted reflections and interruptions during the inspection process. To mitigate any adverse effects, Realtime employed a creative and adaptable solution featuring cross-polarization and specific colored spotlighting.
All of this would be of no use without the proper image resolution to detect the product features of interest. Koya’s product requires varying-sized features to appear clearly within the camera’s field of view and in focus. Considering the drastic contrast in size of the product features of interest directly drove the camera specifications, optics train, and positioning of the camera within the fixture.
Additionally, programming to sequence the sensors and indicators, considering the operator being in the field of view at expected (and unexpected) points during the inspection process was also a challenge. For example, the operator’s hands may occasionally appear in frame, so the programming must be robust enough to ignore the visual data that includes these elements that are not of interest to the inspection process. The team designed extensibility into the inspection fixture design to accommodate future variables or changes in the product features of interest.
Human operator visual inspection is only 80% effective — more challenging inspections are even less effective. Currently, Koya Medical is inspecting their product visually, reviewing up to hundreds of assemblies a day. This AOI system will not only be significantly more effective, but Koya will be able to reduce inspection time from minutes to seconds. Once deployed, in its first week of operation, the new inspection system was able to rectify a 3-month product inspection backlog and increase throughput immediately.
With the successful implementation of this visual inspection fixture in their manufacturing process, Koya immediately realized impressive efficiency and productivity gains. Once duplicated, the implementation of multiple AOI systems installed in their manufacturing facilities will enable Koya to scale their operations and provide life-changing medical devices to even more patients.
Copyright © 2024 The Realtime Group - All Rights Reserved
