An Edmonton entrepreneur is working to develop autonomous tractors with the help of a federal grant.Mojow President Owen Kinch won his grant application under the title "Autonomous Tractor Kit for Enabling Autonomous Farm Implement Operation." The grain farmer spent seven years in SeedMaster’s R&D department, before becoming the first employee of the company’s autonomous vehicle spinoff, DOT Technology Corp.While there, Kinch met his friend and eventual Mojow co-founder, Mojtaba (Moji) Hedayatpour, a systems architect and software developer. While still at DOT, Kinch farmed his own land year-after-year with the company’s autonomous platform. After returning to his own conventional farm, Kinch decided to form Mojow with Hedayatpour.Autonomous Agtech could reduce the agri-food sector’s reliance on increasingly scarce labour. However, existing models generally depend on GPS-driven navigation controllers, which don’t deal well with unplanned obstacles. They therefore require the constant vigilance of a trained human operator.“It’s relatively easy to develop machines that can drive up and down fields independently in straight lines,” Owen explains."But that’s not the entire farming story. Three important bottlenecks or complex challenges must be addressed if agricultural machinery is to operate without human intervention." "The first is driving the headland (untilled land at the ends of fields or near a fence) autonomously without prior mapping." "Second, the equipment must navigate the roads between the fields. And the last big piece is the ability of the tractor to transition seamlessly from road to field using entrances.”Owen says when their technology can do all three, their goal will be achieved.“Our customers’ equipment should drive itself from the yard to the road, access any fields and roads necessary and return to its starting point when the job is done.”The Mojow system employs a series of stereo cameras installed around the vehicle to create a 360-degree unobstructed view of its environment. Everything runs through a proprietary controller called EYEBOX™, which automatically manages frame rates of anywhere between 10 and 30 images per second, depending on the vehicle speed and required job function.EYEBOX™ is a small, rugged, economical sensor suite outfitted with multiple cameras, a GPS, and a powerful computer that processes data in real time. The system collects images automatically, passing them through a deep neural network that classifies each pixel to create or update a digital representation of the whole farm.This digital map feeds the information processed by the autonomous navigation controller. The continuous intake of real-time image data from the peripherals of the tractor assures a high level of relative position accuracy between the vehicle and objects in its working environment.The $2,225,680 project received $682,028 from the Canadian Agri-Food Automation & Intelligence Network (CAAIN). The grant calls for the team to develop technology to detect and respond to the following:all relevant external and internal field boundaries, requiring the development of deep-learning models and software algorithms.all types of field entrance, allowing the EYEBOX™ to employ the most efficient entrance and exit locations.all roadway types, specifically dirt and gravel roads and double track trails.EYEBOX™ is designed to operate ISO 11783-certified farm implements such as air seeders, sprayers and fertilizer spreaders. However, Mojow intends to start with land rolling and heavy harrowing, thereby proving its concept before adding tools that apply product.The platform is flexible enough to convert conventional tractors into autonomous vehicles or be integrated into OEM machinery to enhance functionality. When fully functional, it will reduce a farm’s production costs and reliance on unskilled labour, increasing profitability and productivity.Mojow was founded in 2020, and support from CAAIN has put the company on the verge of introducing a commercially viable product in less than four years.“We’re headquartered in Edmonton mainly to take advantage of the gifted computer-vision, AI, and machine learning professionals trained by the University of Alberta,” Owen notes.“But we wouldn’t have been able to hire so many of the top graduates without CAAIN funding. We’ve also benefited from your network thanks to Garson Law, the program manager assigned to our project. He connected us to a number of companies and individuals whose support has made a big difference.”
