Getting to the Point: Achieving Sub-10cm Location Data to Unlock Next-Gen Autonomous Equipment
April 14, 2022
You’ll no doubt have read about the increasingly autonomous vehicles appearing on our streets. You may have seen adverts for robotic lawnmowers. And you’ve probably heard how autonomous equipment is transforming virtually every industry. There’s huge excitement around the opportunities this type of self-guiding technology promises – not least its ability to liberate humans to spend more time on high-value tasks where machines can’t (yet) compete.
Forecasts suggest the autonomous equipment market is going to be huge. Research by the UK government, for example, predicts the global market for connected and autonomous road vehicles with the highest levels of automation (L3-L5), will be worth USD 890 billion in 2035. Of this, the market for the connected and autonomous technologies that enable these vehicles is forecast to be worth USD 137 billion.
If numbers such as these are going to be realized, autonomous equipment needs to achieve genuine mass-market adoption. And for this to happen, there needs to be a leap forward in the way product manufacturers can access one of the critical enabling technologies: high-precision positioning. This is because devices of the sort we’re talking about need to know, within centimeters, where they are at all times. Moreover, this location information needs to be available reliably and cost-effectively to very large fleets of devices.
Why We Need GNSS Augmentation Data
Location information is typically calculated using data from one of the global navigation satellite systems (GNSS), such as GPS or Galileo. However, using GNSS data alone cannot deliver the accuracy required for high-precision devices. Factors such as satellite clock and orbit errors, signal biases, and ionospheric and tropospheric influences, mean you can only reliably pinpoint device locations to within 2-5m.
To achieve the sub-10cm level of precision needed for next-generation kit, you need to use a GNSS correction or augmentation data service.
These services use a network of fixed ground stations to collect data from GNSS satellites. They calculate the station’s position from the satellite data and compare this to the station’s known location. This enables the service to identify and correct any inaccuracies that equipment close by will be experiencing. This data is then distributed to nearby devices, which use it to refine their own GNSS-based position calculations.
The Challenges with Conventional Augmentation Data Services
Augmentation data services are nothing new, but conventional approaches have limitations. These add significant cost, complexity, and delay for product teams, or can mean the augmentation data service can’t be used for certain applications.
Inability to Scale Sustainably
The first major challenge is around scaling to support large device fleets without breaking the bank. Real-time kinematic (RTK) augmentation services, for example, require bandwidth of around 4.5 kbps and rely on two-way communication between the service and each end device. Precise point positioning (PPP) services, on the other hand, use one-way communication, but still demand between 2.5 and 5 kbps. In both cases, network costs can quickly become prohibitive at scale.
Difficult to Integrate and Manage
The second challenge is around ease of integrating the augmentation data and ongoing fleet management. Some services, for example, use proprietary message formats, while the HTTP-based Networked Transport of RTCM via Internet Protocol (NTRIP) 1.0, used by other services, isn’t natively supported by communication modules.
Using either type of service therefore adds significant and ongoing overhead for engineering and product management teams, which reduces the time they have available for high-value innovation work.
Reliability and Service Coverage
Many of the applications for which sub-10cm accuracy is required are mission-critical, meaning access to the augmentation data service must be fast and reliable. Coverage also needs to be available wherever the device could potentially be used.
Some augmentation data services can take anything from a few minutes to half an hour to achieve an initial sub-10cm location, which won’t be fast enough for many use cases.
Other augmentation services, meanwhile, rely solely on IP-based communication, so will only work where mobile internet access is available.
Elsewhere, accuracy can vary, if the service sends augmentation data calculated solely by a device’s nearest reference station. The further the device is from that station, the less accurate its calculated location reading will be.
Designing a Next-Generation GNSS Augmentation Data Service
To address these challenges, u-blox set out to create a next-generation state space representation (SSR) augmentation data service, with the needs of those designing tomorrow’s autonomous equipment at its core.
Known as PointPerfect, the service typically provides accuracy of between 3 and 6 cm and startup times of 10-30 seconds. It works with u-blox and other commercially available high-precision GNSS modules that support the SPARTN data format (more on this below).
To understand how PointPerfect addresses the challenges today’s product vendors are facing, let’s explore how it’s technically different from other services. This will enable organizations to make an informed decision on which augmentation data service type to use for their products.
Efficiency at Any Scale
To make PointPerfect a viable solution for organizations operating device fleets of any size, its first core principle is efficiency.
Augmentation data messages are sent using the highly efficient Secure Position Augmentation for Real Time Navigation (SPARTN) open data format, over the MQTT messaging protocol, which itself can be significantly more efficient at scale than HTTP. Using MQTT therefore makes it easy for users to leverage augmentation data, whether they’re operating a fleet of a hundred devices, or a fleet of a million.
Communication is via a single outbound broadcast stream to which devices subscribe, as opposed to requiring two-way communication with each connected endpoint. This results in maximum bandwidth requirements of 2.4 kbps and minimizes network transmission costs.
Widespread Availability of Reliable Augmentation Data
Broad and reliable availability is the second core principle. This is a must-have, given many engineers will be designing devices that need to operate in areas without mobile internet coverage. PointPerfect GNSS augmentation data is transmitted through both L-band satellite signals and mobile IP. The development team rigorously tested this satellite capability in some extreme environments (more detailed in the next section). The dual communication channels help align end devices with the PointPerfect service’s 99.9% uptime guarantee, across the geographies where it’s available (currently most of Europe and the contiguous United States, and up to 22 km off coastlines).
Moreover, devices receive augmentation data based on multiple reference stations, for increased positional accuracy.
Self-Service Access/Ease of Use
The third core principle of PointPerfect is to be simple to integrate, and to enable straightforward management of large device fleets.
SPARTN is an open data format, and MQTT messaging protocol is natively supported by most commercial modems. This means engineers don’t need to integrate proprietary client software or build custom modem integrations. Put together, this reduces risk and time-to-market.
In addition, PointPerfect is delivered via an enterprise-grade cloud platform called Thingstream, providing product teams a self-service environment in which to deploy and manage their devices at their convenience. The platform includes a graphical data flow manager to build messages and connections to devices, as well as an API for integrating with other services.
Putting the L-Band Satellite Connection to the Test
With lots of autonomous devices needing to operate in areas with poor or non-existent mobile internet coverage, having a second channel through which to deliver GNSS augmentation data is essential. PointPerfect uses satellite-based L-band communication for this purpose.
Extensive road tests in a variety of challenging locations were performed during the development of the PointPerfect solution to validate the reliability of L-band as a delivery mechanism for this potentially mission-critical data.
Rigorous testing showed that even in some of the most challenging conditions within the PointPerfect coverage area, the augmentation data received via L-band could be successfully used to pinpoint location to within the necessary limits at all times.
The Testing Approach
With PointPerfect designed for the mass market, it was important to establish it would work effectively without the need for highly specialized (and therefore expensive) equipment, such as geodetic receivers. This meant using a standard-performance Tallysman TWA928L automotive antenna on the roof of the vehicle to receive the L-band satellite signals, and the u-blox F9 high-precision GNSS technology platform with multi-band capability.
The key test was to ensure the data received via L-band was continuous, complete, and usable, such that we could calculate our position to within the required levels of precision at all times.
Testing in Challenging Conditions
We planned our routes to take us to areas where satellite signal reception can be challenging. This included going into the Arctic Circle in northern Sweden to test that signal reception would be sufficient when the elevation of the satellites in the sky is very low. At the most northern point on our journey – 66° north of the equator – satellite elevation dropped to just 16°, compared to the 21.8° you get in Helsinki, and 40° in Rome.
We also headed to some of the most extreme conditions found in our coverage area, to ensure the signal strength would be sufficient in these environments. This included the enormous deserts of south-eastern California, the vast mountain ranges in Washington State, and built-up urban areas such as Seattle and San Diego, where high-rise buildings can impact satellite signal reception.
The maps below show two of our test routes.
Despite the challenging conditions, the L-band signal received throughout testing was sufficiently strong to enable reception of the augmentation data service successfully. As a result, we were able to pinpoint our location with horizontal accuracy of less than 10cm, within 2 sigma, or 95% confidence interval throughout the process.
Free Your Engineers to Innovate
Across the industrial, automotive and consumer technology sectors, the next wave of autonomous devices will rely heavily on high-precision positioning information. Those designing tomorrow’s autonomous products need a way of leveraging GNSS augmentation data, without the overheads inherent in conventional services. By removing these challenges, product vendors will liberate their engineering teams to focus on high-value innovation work, reduce time-to-market, slash ongoing maintenance overheads, and enable themselves to offer competitively priced next-generation products and services.