12 Months, 7+ Prototypes, 1 New OKO

An inside look into the countless hours, frustrations, inspirations, and all around passion that went into delivering the OKO product and platform that the AgTech industry sees today.


With the first half of 2021 behind us, ecoation nation is miles away from where we found ourselves just one year prior. In the first half of 2020, the team outlined several improvements to better meet the needs of our customers. Since then, the ecoation’s engineers, data scientists and plant specialists have worked countless hours to bring the next generation of OKO to the market in less than a year. Since launching in December of 2020, OKO sales are up astronomically, communicating much needed confirmation that the team is on track to meet and exceed customer’s expectations. Between then and now, what went into building the new OKO? It’s time to give an inside look into the countless hours, frustrations, inspirations, and all around passion that went into delivering the product and platform that the AgTech industry sees today.


A Year Ago: Challenges to Address, Opportunities to Capture


When bringing an innovative product to market it’s almost guaranteed that unforeseen problems and use cases will appear within the first year. In the case of ecoation’s OKO platform, we wanted the images to provide more detail and an immersive experience to truly act as growers eyes on the ground and machine to work better at even faster speeds. Time is money and receiving insights and information fast is essential to enabling the grower to stay on top of their operation and support their decision making process . Ultimately, these were opportunities to take the product to the next level and introduce an upgraded version based on invaluable customer feedback.


Building the Next Generation: Redesigning From the Ground Up


To make the next generation of OKO a reality, the hardware team conceived and executed a ground-up redesign of the hardware components using all the lessons learned. First, the problem was deconstructed into a set of requirements with priority placed on tackling the fundamental technology development and conflicting challenges. To improve image quality, the team developed an imaging system offering a 360-degree 8K view of the entire crop and various features to capture environmental data including temperature, humidity and CO2 levels at both the top and bottom of the canopy in every row. In an effort to minimise cost for our customers, the new hardware was designed to be fit to the most popular models of existing scout cart by using a modular approach. These modules also allowed for the existence of different hardware packages which enabled each customer to build a data collection system that serves the needs of their unique greenhouse operation.


Initial prototypes consisted of simple 3D printed and hand-made mockups of different approaches that might prove successful. As new technologies were developed in parallel, the performance of the OKO prototypes continued to increase. Roughly 6-months in, the first prototype that integrated all of the major subsystems was created with the intention of demonstrating that the desired performance of each function was feasible. Once proved feasible, successive prototypes quickly optimized each function and the team began to add details, including designing for specific manufacturing and assembly processes. A number of high resolution prototypes were built to undergo full system testing and field trials to gather insights needed to build pre-production prototypes used to provide confidence that all machines built would pass every quality and regulatory test.

“Everything had to be carefully tested and optimized. In many cases we had to write and validate the test methods to ensure they really reflected the customer's needs as well. The whole team really stepped up, understood the customer's needs, and took ownership of the technical solutions they had to deliver to enable the final product. It really is impressive to think we started from a blank sheet of paper just 12 months ago and already have multiple machines in the market, with many customers coming back a month or two later asking for more machines.” - Steve Humpston, Senior Product Design Manager

From a software perspective, a major challenge revolved around enhancing the image quality. Proposals for 3 different hardware cameras (bought & built) were used as initial prototypes and passed to the software engineering team to mock up feasibility demonstrations. After months of trials, the final decision was made to assemble our own camera module, which required bringing in a major partner and purchasing image stitching software. Today, ecoation remains the global exclusive buyer for this innovative software in the AgTech industry. The software was integrated into ecoation’s own camera module which now included 4 lenses plus an extremely powerful edge computing System on a Chip (SOC), resulting in 10x better performance than the previous generation of OKO. To meet the December 2020 launch deadline, the entire device team of engineers rewrote all hardware support code (firmware), re-factored ecoation’s robot operating system (ROS) code to implement two devices on the same OKO machine, reconfigured data uploading and telemetry logic, and more.

“We designed and built all the software from the ground up for a customer hardware product to run a more capable machine in roughly 6-months. By all measures and feedback currently, the machine is more reliable, with a greater number of features, at production quality level currently running in greenhouses.” - Matthew Cox, VP Software Engineering

Building the capability to run machine learning models on the device itself was both necessary and challenging for the software team. The team designed and built a software pipeline that streamed images from the new camera through processing nodes for minimized versions of the machine learning models. This critical technical component ultimately enables the OKO to ‘look’ at the crop and generate valuable metrics immediately on the device. This solution eliminates the need to upload large data files to perform machine learning in the cloud.

“OKO platform with immersive 360 pictures offers a new window into the greenhouses. The opportunities that this data unlocks are limitless. Over the last 12 months we built the hardware and software that enables the technology to run , and we continue building and expanding its capacity ” - Maryam Antikchi, Co-Founder and CTO

Similar posts