LeanPath Weighs In
Choosing Between Open-source and Commercial OS

Most of us designing embedded systems have a pretty good idea of the parameters involved in choosing an OS for our device.  We weigh the various business and technical concerns – open-source versus commercial, royalty-based versus one-time license fee, supported versus roll-your own.  What happens when a startup company wanting to apply embedded technology to commonplace, real-world problems in traditionally low-tech areas needs to choose an OS for their product?

LeanPath is a Portland, Oregon-based company that wanted to develop an intelligent system to attack the problem of food waste in commercial and institutional environments.  In the food service, restaurant, and hospitality industry, a great deal of food is wasted due to the lack of intelligent monitoring and management of production and waste.  They reasoned that a system that tracked what food service operations were throwing away would have the basic data needed to improve the planning efficiency of most operations.  Once the data was collected, analysis tools would help management make better-informed decisions.

While Mom may have been able to accurately guess how much you and your siblings would scarf down at a typical dinner, the problem a hospital kitchen faces in deciding how much of which food items to prepare is considerably more complex.  Also, it’s easy for Mom to check to see how much wasn’t eaten, adjust the recipe for next time, and recycle the leftovers into a …um… delicious casserole for the next night.  In a large-scale food production operation, much of the waste is typically unmonitored and goes directly into landfills.

Leanpath reasoned that a system could be developed that would allow food service operations to easily and accurately monitor food waste and use the information to optimize their menu planning.  By accurately tracking what food was discarded at the end of the day, food service managers could save money and significantly reduce waste.  The challenge was producing a system that could easily be operated by unskilled employees with minimal training and that could stand up to the harsh environment of a commercial kitchen.  The system also needed to be robust and require minimal maintenance and system management.

The system is comprised of two major components: a software application called “ValuWaste Advantage” that runs on a conventional desktop machine, and one or more “ValuWaste Tracker” systems, which are single-board computers embedded in stainless steel scales.   Foodservice workers use the Tracker scales to weigh and categorize pre-consumer food waste.  Data is passed from these systems back to the central database management system using either USB thumb drives or via a network connection. 

Leanpath chose the Applied Data Systems (ADS) BitsyX as the hardware platform for their Tracker product.  BitsyX is a single-board computer based on the 32-bit Intel PXA255, with an SA-1111 StrongARM companion chip that provides USB host and support for PCMCIA and Compact Flash.  With minimal effort, the company was able to interface the single-board computer with the scale.  Next they needed to select an embedded operating system for the Tracker.

“We initially thought we’d choose a variant of Linux,” says LeanPath CTO Stephen Rogers.  “At first we assumed that the no-royalty aspect of open-source would pull us in that direction.  However, as we analyzed our requirements, we found it would take considerably more development effort – and that we’d have to take on a large part of the OS development responsibility just to get our project off the ground.  We evaluated a number of options, including Linux and Windows Embedded CE, and found that, with our small team size and tight development budget, Windows CE was a better choice.”

The team had no prior Windows CE experience.  With two engineers working on the project, they decided on C# as an implementation language.  With both engineers learning Windows CE and C# for the first time, the entire project was still completed in five months.  “The emulator worked great for us,” Rogers continues.  “We had simultaneous tracks going on hardware and software development, and the emulator let us get a lot of software development done earlier in the process.”

The Leanpath embedded software project faced some unexpected obstacles.  One developer became seriously ill during the project, recovered, and rejoined.  The requirements evolved as the project progressed (as is often the case in startup environments), and the team had a “minimal” and a “stretch” set of functionality bracketing their completion target.  “We first designed the system as a non-networked application,” Rogers explains.  “Most kitchens don’t have Ethernet running around.  Later, we added networking capability.”

The continued evolution of the product was another happy aspect of the team’s selection of a commercial OS.  As with the early evolution of any new product, new hardware configurations had to be supported, new features were required, and bugs were uncovered by customer experience.  The robust development tools and continued support of a commercial OS vendor allowed the team to focus on the value-added aspects of their application, rather than wasting time wrestling with OS concerns.

One of the places that product evolution took the team was onto cruise ships.  In a major business win, Royal Caribbean International decided to use the ValuWaste system on board the world’s largest cruise ship.  With over 18,000 meals per day, the potential for waste reduction was enormous.  Of course, as with any engineering task, the devil was in the details.  With the first installations, the field teams noticed that the movement of the ship made the scale unstable.  The system couldn’t get an accurate reading.  Almost immediately, the software team had a patch – they added code to stabilize the readings in software, shipped the patch, and the system was up and working correctly on the same day. 

One of the benefits of building an embedded product around commercial, standardized components such as the ADS board and the Microsoft OS is the ongoing improvement of much of your system without additional engineering investment on your part.  The LeanPath team is already evaluating the upgrade to a later version of Windows CE, and they’ve already absorbed several hardware improvements through ADS.

One of the biggest lessons from the LeanPath experience is the engineering leverage you can gain by avoiding the NIH (“not invented here”) syndrome.  LeanPath took advantage of pre-engineered hardware and software components so that they could focus their engineering efforts on the parts of the system where they had specific expertise.  Often, a false economy results from the make-versus-buy decision process, particularly when ongoing maintenance and enhancement are factored in.

Kevin Morris, Embedded Technology Journal

May 8, 2007

[back to top]