Farms are greatly improving productivity by applying automation to even such fundamentals as planting and weeding. Some methods of automating those processes require precise positioning, however. One technique is to deploy multiple types of sensors, and use the combined data collected to improve positional accuracy. Another is to start with an available positional signal, which generally tend to be accurate to meters, and then apply one or more correction techniques to improve its accuracy.
The latter approach is the one taken by Point One Navigation. The company relies on GNSS, which it improves by means of RTK (Real-Time Kinematic) solutions to provide positioning that can be accurate to within centimeters, explained CEO Aaron Nathan. We spoke with Nathan and Mike Coletta, Point One’s vice president of sales, to get details about their company’s solutions.
Point One Navigation
“By taking the GNSS signal, which has maybe five meters of accuracy or a little better, and applying our corrections, we have been able to get down to five centimeters of accuracy,” said Nathan.
The company has built, from the ground up, a proprietary network of beacon stations with features that including anti-jamming, interference mitigation, security, and integrity monitoring. It also supports multiple correction types and all GNSS constellations. The network is called Polaris.
Spacing beacons roughly 100 kilometers apart covering, Polaris covers most of the the continental US (see map).
“We first looked around in the US and decided we could do better by deploying more modern equipment and improving the convergence time to get down to that ultra-precise level. We are expanding and moving into Europe now as well,” said Nathan.
Point One is capable of determining position within 5 cm, but accuracy depends on part on environmental conditions, including the presence of objects that might block the system’s signals. The company’s goal is to consistently provide accuracy of 10 centimeters. In fact, the company’s name is based on the goal — “point-one meter.”
Even though 10 centimeters accuracy is commonly accepted in automotive or robotics applications, smart agriculture sometimes requires better than five centimeters of accuracy. Planting, for instance, requires high precision for the location of seeds.
Station density and management are prerequisites for that, the company argues. According to the Point One execs, “Some farmers are using local RTK stations and 90% of the time that works fine. However, if a local base station has an issue or it is moved, the whole operation can be interrupted. What we offer is a network where stations are not standalone, but fully managed by our cloud infrastructure.”
The company does 24/7 monitoring to ensure all stations are correctly operating within relevant parameters, such as battery voltage and state of charge. Dedicated sensors are also able to detect tampering, notifying in real-time if someone is trying to take the unit down.
Besides the hardware, which is all developed internally, another relevant aspect of Point One’s solution is the software. Mainly written in C++ language, the software has been extensively tested and is compliant to ISO 26262, a standard employed by most automotive companies for the development of functionally safe code.
The sensor fusion software, named FusionEngine, has been tested over weeks, or even months, of driving time. This allowed Point One to build an ultra-high-fidelity record in playback, which is useful for running regression tests after any modification to the source code. “One of the things that we excel in is sensor fusion, meaning our algorithms are able to manage redundancy of sensors to improve the position accuracy,” Nathan said.
FusionEngine is a multi-frequency GNSS/INS positioning engine featuring dead reckoning capabilities. It combines data from multiple sources to achieve high accuracy in more places and resilience to extended blockages.
Given that LiDAR sensors are still relatively expensive, Point One makes use of commonly available and low-cost sensors such as GNSS receivers, IMUs (inertial measurement units), vehicle network signals and cameras. FusionEngine adapts to a wide range of environments such as urban, highway, suburban, multi-story structures, tunnels, hills and altitude performance.
Point One recently announced a collaboration with Quectel Wireless Solutions, a global supplier of IOT modules and antennas. The integration of FusionEngine into Quectel’s module aims to solve positioning issues, especially in urban and suburban environments. GNSS with RTK and dead reckoning can accurately distinguish levels of multi-story structures, which the photo below aims to demonstrate. Moreover, dead reckoning helps to overcome degraded performance from outages in the correction service.
“Through this partnership, we have been able to integrate our core product (FusionEngine) into their LG69T platform, which is a very major accomplishmen,” Nathan said.
The LG69T-AM GNSS module, based on the STMicroelectronics’ TeseoV dual-band L1/L5 positioning receiver, features 80 tracking and 4 fast acquisition channels compatible with GPS, GLONASS, Galileo, BeiDou, QZSS and NAVIC. “The integration of FusionEngine on the LG69T platform brings us not only into automotive, but into many other vertical markets, such as micro-mobility and smart agriculture,” Nathan said.
Last December, Point One also announced a partnership with Ligado Networks, who will provide pervasive, high-performance satellite capacity to support Point One’s Polaris GNSS correction service. The partnership will initially provide L-Band service to electric vehicles with advanced driver assistance systems (ADAS) and will expand to additional sectors in the coming years. “By having that redundancy relying on L-Band satellites, we can provide a backup delivery mechanism, which is extremely important in automotive. Actually, in some applications, that could become the primary method of delivering the corrections to the asset as well,” Nathan said.