De-risking Your Vision-System Product Roadmap

You want to embed a camera or vision system into your product to drive more value? What’s the best approach?

Companies across a range of industries (automotive, home automation, drones, surveillance, and so on) are adding vision to their products. It’s no longer just about the image, but also the data and insight that machine intelligence can derive from the image to gather data, prompt an action, or support a decision. Embedded vision has become the key component of the Internet of Things (IoT), to add relevance and intelligence to connected devices and appliances at the network’s edge.

These product companies want the reliability and robustness typical of industrial machine vision at a consumer price point. Concerns about lack of control over costs, ownership of IP, and quality assurance leave them reluctant to outsource this to a third party.

Instead, they attempt to build this vision capability in-house. But are they setting themselves up for failure?

It’s Not as Easy as It Looks

It’s true that that the cost and technical complexity of designing and building a vision system had fallen in recent years. We see this in the rise of CMOS (complementary metal-oxide semiconductor) sensors, as they improve in quality and functionality with the constant evolution of mobile device technology versus CCD (charge-coupled device) sensors that require far more tedious calibration.

New chip designs allow for more powerful onboard image processing and camera performance with off-the-shelf components. Common data-interface standards have reduced the need for custom software development. Component makers are also offering kits that accelerate the development process and shorten time-to-market. These kits, however, often feature “middle of the road” components meant to be all-purpose, which means they may not be ideal for a specific application.

Some component and design costs may have dropped in recent years, but the complexities of configuration and tuning aren’t easing up. In fact, they have arguably increased, as image sensors decrease in size to fit smaller camera module form factors.

Who’s Going to Build It?

Technical specifications and considerations are only part of the equation. Components must be sourced, integrated, and tuned appropriately to the intended application. The module must be designed to be manufactured at scale, at a cost point, and to a reliable standard of quality that will support a product company’s business objectives.

This leaves the product company having to source components from resellers, negotiate with integrators, and engage with contract manufacturers, the majority of which are located in China.

The product company took this project “in-house” due to cost, IP, and quality concerns. But without prior experience in the vision-system design and manufacturing industry, the company has instead left itself trying to navigate unknown waters. It must attempt to create and manage supply-chain relationships when it has little to no experience dealing with overseas parties.  

Many product companies come to iENSO after they have attempted to go it alone. We hear the horror stories about bargain deals for what proved to be faulty and sub-standard components, quality-assurance issues with contract manufacturers, cost overruns, and delays getting to market. While these product companies were dealing with these issues, their competitors were winning market share or catching up.

A product company that doesn’t already have the technical expertise and supply-chain relationships to confidently handle its embedded vision needs in-house needs to find the right collaboration partner—ideally, one that’s based in North America, but with deep roots in Asia.

10 Questions to Find the Right Partner

1. Does this collaboration partner have the right mix of standard and full custom design services to address your every technical and performance requirement, as well as the ones your team hasn’t thought of yet that are specific to your application?
2. Does it have the flexibility to cost-effectively design for your intended scale of production run—low, medium, or high?
3. Does it have the relationships that allow you to integrate the best technologies and innovation into your product roadmap?
4. Does it understand the full value and potential of edge AI and innovative ways to align vision and AI?
5. Does it have the ability to protect your IP?
6. Does it consider data security and privacy as part of your product design? Data truly is the new oil. Any connected device yields valuable data for product improvement or insight into consumer behavior. But this data, and the privacy of the individual, must be protected. The California Consumer Privacy Act (CCPA) is just one example of the regulatory compliance burden that may apply to your product in today’s market.
7. Does it offer expert manufacturability methodology and factory-setup services that will ensure you can quickly scale from prototype to mass production without sacrificing product quality or performance?
8. Does it have those boots on the ground in Asia to manage deadlines and quality assurance with contract manufacturers?
9. Does it assist with your product roadmap and go-to-market strategy so that you can remain competitive and focused on your core business?
10. And, most importantly, does it have a roster of happy clients with businesses comparable to yours?

Regardless of your embedded-vision expertise or lack thereof, and regardless of what you have already attempted and where you may be stuck on your product roadmap and go-to-market strategy, don’t feel obligated to outsource to a distant third party. Instead, look for a collaboration partner that’s accountable, capable, and proven.

Sebastien Dignard is President of iENSO.