| Lynn
Recently, positioning sensor supplier Fixposition officially launched its four-year-old global high-precision positioning product Vision-RTK 2. This is a new generation of multi-source fusion positioning sensors, aiming at the needs of unmanned vehicles, drones, robots, etc. for full-scene, high-precision global positioning, etc., to help the user platform achieve safe and efficient automation in the application.
Vision-RTK 2 deeply integrates computer vision, RTK-GNSS, inertial sensor (IMU) and wheeled odometer data, combined with machine learning algorithms, even in harsh scenarios such as GNSS (Global Navigation Satellite System) signals are blocked, shielded, and subjected to multipath interference such as reflection refraction, it can output continuous, real-time, and reliable centimeter-to-decimeter-level high-precision global positioning.
Through the test video released by non-general navigation technology, it can be seen that the unmanned vehicle equipped with Vision-RTK 2 runs through the square, pedestrian passage and underground parking lot with complex road conditions for many times, and under the frequent indoor and outdoor scene conversion, the real-time positioning vehicle trajectory and the actual route are highly consistent and the trajectory is smooth, perfectly avoiding the position shift caused by multipath, weak signal, insufficient number of satellites, etc.
After actual measurement, the horizontal and elevation positioning accuracy of Vision-RTK 2 reaches 1.0cm + 1ppm under the RTK FIX state, and the decimeter-level positioning accuracy can also be achieved in the urban "canyon". Even in the absence of a navigation satellite signal indoors, position drift is controlled from an error of 0.5% (with wheel tachometer data) to 1.5% (without wheel tachometer). It is worth mentioning that compared to the position drift of conventional inertial navigation products (RTK+IMU+WO), which accumulates over time, the drift of Vision-RTK 2 is only related to the distance traveled indoors. The Vision-RTK 2 is therefore particularly suitable for low-speed autonomous driving platforms that require operation in complex environments with poor satellite signals.
The products released this time are optimized for small and medium size, medium and low speed (less than 100km/h) and ultra-low speed ground robot platform, especially suitable for outdoor logistics robots, unmanned delivery vehicles, outdoor automatic cleaning robots, agricultural robots, automatic lawn mowers, etc.
Compared with other similar products on the market, Vision-RTK 2 wins with high cost performance, and realizes the global positioning of continuous output of centimeters to decimeters in complex and harsh environments at a reasonable price, and there is room for further optimization of the price after large-scale mass production.
Behind the excellent product
The vision-RTK 2 has such excellent performance, which is inseparable from the company's strong product development and market analysis team.
In August 2017, two pioneering teams of scholars from ETH Zurich founded Fixposition AG, which has more than 10 years of scientific and practical experience in satellite navigation and computer vision technology. ETH Zurich is Einstein's alma mater, a world-renowned top research university, with world-leading laboratories and scientific research capabilities in the fields of robotics, unmanned driving, positioning and navigation, and enjoys the reputation of "the first school in Europe". After the establishment of Feipu Navigation Technology, it was quickly selected to join the Swiss Business Incubation Center of the European Space Agency. Since then, the company's team has continued to grow, attracting many senior engineers and experts from positioning and navigation leaders to join.
Su Zhen, co-founder and CEO of Feipu Navigation Technology, mentioned that during the doctoral research at ETH Zurich, the core of his scientific research project at the beginning was to improve the positioning accuracy of consumer-grade GNSS (such as the positioning chip used in mobile phones) from a few meters to a few centimeters. In the last two years of the doctorate, coinciding with the rapid development of global automatic driving technology, Su Zhenzhong realized that high-precision positioning will have a broad market demand in the field of automatic driving, and he began to start a business from that time, committed to creating a high-precision positioning product that can help the large-scale landing of automatic driving. Another co-founder and CTO, Lukas Meier, is also a pioneer scholar at Swissotech in Zurich, focusing on computer vision and robotics. By innovatively combining vision with satellite navigation, non-general navigation has found a technical route that is different from conventional.
In December 2019, the company completed a seed round led by True Ventures, a top venture capital in Silicon Valley. At the same time, Daniel Ammann, co-founder of u-blox, a leading navigation company, also participated in the follow-up investment and joined the company's board of directors.
With the product mass production ready, in October 2021, the company received another seed + round of financing from old shareholders. At the same time, it has also attracted many excellent new investment institutions at home and abroad, including Dr. Lu Qi's miracle achievements.
"We have been deeply involved in visual positioning, high-precision satellite navigation positioning, and multi-sensor data deep integration technology for many years. In October 2020, the first generation of Vision-RTK engineering tester was released and won wide acclaim in the market and customers. While determining that the market has a strong demand for this technology, we also humbly accept a lot of customer opinions and feedback, and are more aware of customer needs. With these gains, after a year of polishing and optimization, we officially released the upgraded mass production version at the end of March this year, that is, Vision-RTK 2. Su Zhenzhong introduced that this is a very breakthrough, cost-effective high-precision positioning products.
Vision-RTK
Vision-RTK 2
Su Zhenzhong said that the current unmanned navigation and positioning methods are diverse, but the most widely used outdoor is satellite navigation and positioning. RTK-GNSS can obtain centimeter-level positioning accuracy based on carrier differential positioning technology, but it can only be achieved in open areas where navigation satellite signals are well received, and real-time high-precision output cannot be achieved once entering complex scenes, such as areas under tunnels and viaducts, where the signal is blocked and weakened. In this case, purely based on satellite navigation technology, it is even impossible to obtain usable positioning; in the urban "canyon", due to the multi-path effect of high-rise buildings and the multi-path effect of reflecting satellite signals, the positioning accuracy will be reduced to several meters or even tens of meters; and the pure visual navigation positioning or lidar based on the camera, although it can be well positioned relatively, but in bad weather such as rain, snow, fog, etc., its performance cannot be guaranteed.
Therefore, no single positioning technology can meet the high-precision positioning requirements of the whole scene. This pain point is one of the reasons that hinder the large-scale landing of autonomous mobile robot applications.
In response to this pain point, Non-General Navigation Technology launched Vision-RTK. Through the deep integration of computer vision and high-precision satellite navigation and positioning, strong complementarity allows unmanned vehicles and autonomous robots to maintain high-precision and high-stability positioning in various complex environments.
As the second generation of the newly upgraded product, the Vision-RTK 2 has greatly improved the performance of the product and has a richer interface to meet the needs of more customers. Su Zhenzhong specially introduced that Vision-RTK 2 is not limited to the application of certain scenarios and certain types of products, its core value is to expand the global positioning based on GNSS to more scenarios to help customers optimize autonomous driving solutions and reduce the cost of their positioning and navigation, and ultimately expand the application scope of customer products. Therefore, in addition to vision and satellite navigation technology, Vision-RTK 2 also integrates inertial sensors, wheeled odometers and integrated machine learning algorithms, this innovative fusion method can not only solve the problem of GNSS signals being blocked and reflected, but also avoid the interference of pure visual odometer positioning by strong light changes. Whether it is day or night, or even when driving from the underground garage to the open area, the Vision-RTK2 can achieve real-time stable and high-precision positioning output.
At present, Vision-RTK 2 is mainly used in low-speed unmanned vehicles, vehicles in some special industries, unmanned agricultural machines, or autonomous robots such as drones and lawn mowers, which have clearer and stricter requirements for positioning in various scenarios.
Partner's unmanned vehicle platform
For example, unmanned delivery, unmanned patrol vehicles, the use of Vision-RTK 2 can help vehicles pass through the urban "canyon", enter and exit the garage is not affected by signal reflection, light changes, to obtain accurate location.
The customer's mower robot
Such as lawn mowers, even if the robot frequently passes through the shade, the Vision-RTK 2 multi-source sensor deep integration technology can ensure that the robot achieves centimeter-level positioning and operates normally within the boundary range. In addition, it can avoid the problems of time-consuming and costly maintenance faced in the later stage of the boundary demarcation operation with buried lines, and truly realize the automatic mowing of the borderless line.
Technology leads, creates the future
“The best way to predict the future is to create it.” In Su Zhenzhong's view, although the non-general navigation technology is still a very small team at this stage, it is gathered by a group of people who believe in their dreams and create the future.
Group photo of the non-general navigation technology team
In order to build a ubiquitous high-precision positioning and make automatic driving simpler and more reliable, the company's team has been deeply involved in related technologies for many years and developed Vision-RTK products, and quickly iterated to the second-generation mass production version, providing better positioning solutions for more autonomous vehicles and autonomous robots. This is a milestone breakthrough for non-general navigation technology; it also injects a new technological force into the unmanned driving industry.
At present, Non-General Navigation Technology has nearly 30 customers in China, the United States, Germany, France, the Netherlands, the United Arab Emirates, South Korea and other countries around the world, and has signed a framework volume order contract with domestic AgileX, PIX Mobile Space (PIX Moving), signed a memorandum of cooperation with Neolix, and carried out in-depth cooperation with 2 head mowing robot companies.
Su Zhenzhong said that the company hopes to take agriculture, logistics and lawn mowing robots as the entry point, and finally realize the large-scale application of autonomous robots to meet the just needs of precision agriculture, smart logistics, smart cities and industry 4.0. At present, the customers of non-general navigation are mainly concentrated in China, the United States and Europe. Among them, the degree of mechanization of agriculture and forestry in Europe and the United States is high, and the stock market of agricultural machines and lawn mowing robots is huge, and the non-general navigation technology is committed to meeting the demand for high-precision positioning technology in the automation upgrade of their next-generation products.
China's agricultural mechanization and automation are developing rapidly. Coupled with the support of national policies in recent years, the rapid growth of automated agricultural machinery applications has brought about a large increase in the demand for high-precision positioning, and the growth space of lawn mowers and agricultural robots is very huge. The company focuses on incremental markets.
In addition, as one of the members of the low-speed unmanned driving industry alliance, non-general navigation technology is particularly optimistic about the development of China's low-speed unmanned driving industry. Su Zhenzhong believes that at present, domestic unmanned driving technology has great advantages in both scientific research investment and the number of start-up companies, and the overall development speed and scale are ahead of other regions. The low-speed scene can be commercialized as soon as possible, including unmanned distribution, unmanned patrol, unmanned retail, unmanned sanitation, etc., and its incremental market continues to expand.
Wang Daihang, sales director of Feipu Navigation Technology China, has more than 15 years of sales and R&D experience in the electronics industry, and was once the regional sales manager of u-blox. He introduced that for the Chinese market, non-general navigation technology at the end of 2021 opened the Vision-RTK 2 early access plan, the product has been tested in the country for tens of thousands of kilometers, and synchronously in a number of customers' driverless vehicles, autonomous robots on the initial integration and testing, collected a large number of typical customer test scenario data, to solve many customers are currently facing difficulties, while the R & D team also continues to optimize product performance based on these data. After the official release of the product, the company promoted the product to more customers to test the product, and strived to enter the supply chain of more unmanned vehicle and robot manufacturers that needed high-precision positioning solutions.
It is reported that non-general navigation technology currently masters the four core technologies of computer vision positioning, high-precision satellite navigation positioning, deep integration of multi-source sensors, and data fusion based on machine learning. In the future, the company hopes to not only be limited to the iterative upgrade of high-precision positioning technology, but also consider the modularization and softwareization of products to provide more cost-effective products for the automatic driving industry. At the same time, expand the application of technology in the cloud platform, expand the support for the positioning and navigation of pure indoor robots, and further optimize the navigation and positioning capabilities of products with stronger computing power in the cloud and combined with machine learning.
(WeChat search for "low-speed driverless industry alliance" or "wrdrive", pay attention to the public account at any time to learn more information.) ）