The Hatch team successfully completed a custom Android Advertising Display project – from design to delivery. Here's how we overcame some challenges to get it done.
June 21, 2019
Design electronics, firmware, and casing of a custom Android Advertising Display to facilitate the unique functions of our client’s app. The displays are installed on top of supermarket shelves, usually at the check out aisle.
The display shows advertisements for products that sit on the shelf below it with the intention of increasing sales for the advertisers.
In addition, the app uses images from a wide-angle front camera to recognize when people are looking at the advertisement to provide tangible impression data back to the brands which run ads on the display.
Lastly, the front and back cameras, designed at a 15-degree downward angle, scan inventory of products on the shelves directly in front of and behind the display and, through a back-end machine learning algorithm, notify the supermarket in real time to restock products that have sold out.
The Android Advertising Display provides dynamic solutions for targeted and tracked promotion of consumer goods at the critical juncture when customers make buying decisions.
Additionally, the display helps supermarkets reduce lost sales by instantaneously notifying them when products are out of stock on their shelves.
The displays can also show community service announcements or other information besides advertising to enrich the shopping experience and store environment.
Hardware and Mechanical Requirements
Connects to the internet with Wi-Fi (no 3G/4G telecom)
No Touch Panel
11.6” HD Screen
Wide Angle Camera Lens
Industrial design to support 24*7 Use
Small Backup Battery (in case power goes out)
Fixed Focus 5mp Front and Back Cameras
Custom Android firmware engineering
Industrial Design (design of external case)
Mechanical Engineering (design of inside case)
Supply Chain selection and management
Creating CNC prototypes
Tooling for device case
Manufacturing process optimization and management
Create finished product testing procedure
Behind the Smile – Getting it Done
This product is now on its second version.
Reviewing the events of developments from version one and two will highlight how important early decisions dictate the course and outcome of getting a product made based on the contrasting results of developing the two versions.
The initial version took over a year from starting development to finishing mass production for no good reason.
Designing the first version took much longer than it should have because of complications with electronics design and sourcing.
These hassles served as a valuable lesson in highlighting important considerations during planning and before commencing design of a product. The main problems faced during the first development revolved around choosing a CPU which wasn’t optimal for 2 specific design requirements.
Firstly, the industry started moving away from using a MIPI connection type for the 11.6” screen to using an EDP connection type.
Secondly, the wide angle lens added complications to image processing. Hatch choose to build the first generation product around the AllWinner Tech A64, a popular and stable quad core 64 bit Android CPU.
While this CPU gained design wins in standard consumer devices it doesn’t have native support for the EDP connection and it doesn’t have an internal ISP (image signal processor), which serves to improve image quality.
A fundamental oversight on Hatch’s part was designing the PCBA before confirming compatibility with other components (something we usually do), having assumed that the standard nature of the device wouldn’t lead to any surprises.
To resolve screen incompatibility the choices were adding a MIPI to EDP conversion IC to the PCBA or sourcing hard to find MIPI screens (since all new 11.6” screens used EDP).
Hatch decided to look for 11.6” MIPI screens instead of undergoing the long and costly process of redesigning the PCBA to add the conversion IC.
Since it would have been very costly and slow (due to the low manufacturing volume) to get custom screens made with the MIPI connection we had to source stock 11.6” MIPI screens.
It was difficult to find stocks of new screens and most suppliers wouldn’t provide any guarantee on quality of their stocks.
We had to be careful to avoid buying second-hand screens (being sold as new) since they had too many quality problems which would make manufacturing difficult and increase the chances of products breaking after shipping.
After Hatch found a reliable source for factory-sealed stocks of 11.6” screens with MIPI connections the pressure eased a bit, but we knew that finding screens for future production runs wouldn’t get any easier.
The second issue was that the special small size wide-angle camera lens used in the advertising display required better image processing than standard cameras.
Because the A64 doesn’t have an internal ISP we had to choose from a limited selection of, costly and mostly outdated, cameras which have an ISP built into the camera.
Ultimately the camera was a little more expensive and image quality lower than we would have liked, but we found a model good enough to perform the job required.
The client was satisfied with the product after all the extra work making the best screen and camera results possible, but getting the product to this point required way too much time and hassle.
After battling these issues for a few mass production runs, Hatch offered to redesign the product for free.
The problem originated from the short-sighted product architecture Hatch suggested and most of the burden to overcome the problems were borne by Hatch so, as long as the client wanted to continue ordering this product, redeveloping (free of charge to the client) seemed like the best solution for everyone.
The second development took only 2-3 months to have a first working prototype and under 5 months until the first trial production was finished.
This second development went quickly because the same case was used from the initial version with only minor modifications and the experience from the first version ensured we resolved key issues from the beginning.
How it Happened
In spite of being a new company, this client impressed us with the professionalism and background of their management team and advisers, respectively.
Their business model also made sense, inspiring additional reason to believe in their potential.
Before finding Hatch this client worked with another company and felt the pain that often comes as a result.
The hardship of this experience is what my friend Liam Casey calls becoming ‘China Ready’. China Ready clients are the most appreciative and easiest to work with.
For distribution, our client partnered with an established company that sells shelving to supermarkets to add ‘smarts’ to their traditional shelf product line.
By adding the Android Advertising Display, the shelving not only holds products but also generates unique tangible value to the product brands and supermarket.
In general tech companies that use new technology to improve on proven businesses, rather than creating something totally new, have more appeal (from a business success standpoint) than tech companies that try to create demand in unproven categories.
With faith in the business model and the people running the business, Hatch accepted the opportunity to handle the full project from design to delivery.
On most projects, Hatch requests that the client works directly with an industrial designer to make the external case design, but this client said they wanted Hatch to do the external case design for them and they wouldn’t be picky about little details.
The hesitation we have with offering design services is that it’s usually more efficient for the client to work directly with the designer rather than through Hatch and sometimes cheaper for the client as well.
If the client really isn’t sensitive about the casing design and the design is simple (or their designer fails to do a good job resulting in delays to the project) then we may accept the responsibility.
Manufacturing, especially on the first version, required untold hours of on-site quality control to run extensive testing on the 2 cameras of every unit.
Photos were taken from both cameras of each unit to ensure clarity and color accuracy. The cameras that failed were sent for re-calibration.
Hatch staff and its owner personally administered testing of the cameras for this critical task during normal working hours, holidays, weekends, and until late at night to ensure the best results as quickly as possible.
Sample batches were taken from trial and mass production runs to put in heat chambers for stress testing to ensure the devices (and more specifically the screens) would continue to operate on hot days when the air conditioner breaks and the device is left on for 12 hours in a supermarket.
Another special test done on these products was to age test them for 72 hours straight, rather than the standard 4-5 hours for consumer devices, since the client’s application called for the devices to run all day long, day after day.
Since the factory wasn’t willing to do this, Hatch put an age testing rack into our downtown office where about 100 devices at a time ran straight for 3 days.
Devices were shipped back and forth between the Hatch office and the factory using private drivers.
A lot of the extra testing Hatch did happened with the first version as a result of the non-ideal situation with the camera and screen mentioned above.
When bad things happen, understanding and controlling the risk prevent bigger problems.
Eliminating the added burden Hatch had to deal with as a result of controlling the risk was one of the driving factors that led to creating version 2 of the product.
We’ve seen many manufacturing companies that don’t realize the value of proactively prioritizing the long term success of the client or don’t value their own reputation enough to spend significant resources on extra testing or redevelopment.
The transparency and efforts that go into our projects make Hatch different.
The first version performed well enough that the client wanted to continue reordering it.
This validated the extra efforts Hatch put into overcoming the challenges with the screen and camera, but more orders simply meant more pain; opposite of the typical and desired result.
With the second development finished and approved we are confident in the product quality and manufacturing process efficiency moving forward.
Hatch is excited about the future for our client and their Android Advertising Display business.