June 20, 2017
Executive Summary: Investment in autonomous vehicles will reduce the cost of dual-frequency GNSS receivers by a factor of ten within the next five years.
In a previous article we argued that the price of dual-frequency satellite receivers is poised to fall precipitously with the advent of new satellite navigation signals and new markets. We discussed the dynamics driving this process and its implications for the democratization of land surveying, but we did not quantify this price reduction or say when these new, cheaper devices would be available. In this article we address those questions in detail by looking at the projections that government agencies, think tanks, and market analysts have made regarding the development of dual-frequency satellite receivers for the transportation market. There is a widespread expectation that autonomous vehicles will become the first mass market application for dual-frequency GNSS receivers. The vehicle market is best positioned to adopt this technology both because it has the most pressing need and because it is not limited by the energy consumption, processing power, and miniaturization concerns that restrict the use of dual-frequency in drones, wearables, and cell phones. We anticipate that the mass production of dual-frequency receivers for the vehicle market will reduce their price by a factor of ten within the next five years.
The Race Towards Automation
There is no consensus on when fully autonomous vehicles will become common. A May 2017 report from the independent tech think tank RethinkX entitled Rethinking Transportation 2020-2030 projects an exponential adoption curve that will result in 95% of US passenger miles being served by on-demand autonomous vehicles by 2030. Their conclusion, premised on economic feedback loops making individually owned cars prohibitively expensive compared to Transportation-as-a-Service, is far more optimistic than most other estimates. Market reports from business consultancies, while more cautious, still project rapid growth. PwC said in its 2015 Connected Car Study [PDF] that “fully autonomous long-range driving at highway speeds [is] expected between 2020 and 2025.” Mckinsey’s Automotive Revolution - Perspective towards 2030 [PDF], concludes that anywhere between 15 and 50 percent of cars will be highly autonomous in 2030, depending on the degree of market disruption.
Car manufacturers, tech companies, and ride-sharing services such as General Motors, Renault-Nissan, Daimler, Volkswagen, Waymo, and Tesla are all working towards fully autonomous vehicles. Uber and Ford have announced plans to put highly or fully autonomous cars on the road in high volume by 2021. As a result, enabling technologies are receiving significant levels of investment, both because they improve human-driven connected cars and because they are a prerequisite for autonomous vehicles.
This is particularly true of technologies like dual-frequency GNSS that increase safety. In the words of Craig Giffi, leader of Deloitte LLP’s US automotive practice: “to win consumers’ trust, automakers will need to integrate limited self-driving and advanced safety features into new product offerings steadily over time to introduce people to the technology, demonstrate the improvement in vehicle safety and develop a proven track record.” Autonomy is simultaneously appealing and frightening to consumers, who are enthusiastic about the potential of Transportation-as-a-Service to reduce travel costs, pollution, and urban congestion, but will be reluctant to adopt it until they are absolutely convinced it is safe. Autonomous vehicles have the potential to greatly reduce road fatalities, but consumers fear the loss of control and are likely to judge them by a more critical standard than human-operated vehicles. More people are afraid of commercial air travel than of driving despite the fact that flying is statistically much safer. The confidence of regulators and investors is as critical as the confidence of potential users. PwC notes that in the scramble to control the emerging connected and autonomous car market “No one will win ... if security concerns undermine consumers’ trust in connected car technology.”
Why is Dual-Frequency Important for Autonomous Vehicle Safety?
Unlike the current generation of in-dash and phone-based car navigation systems, which only use a single frequency, GNSS systems used for autonomous vehicle navigation must employ at least two frequencies to achieve the highest safety standards, which mandate higher accuracy and signal integrity than single-frequency can provide. As the European GNSS Agency noted succinctly in its 2016 GNSS User Technology Report [PDF]: “For safety-critical applications, where redundancy and resistance to jamming is important, dual-frequency (L1/E1 + L5/E5) is undoubtedly the best choice.” The greater integrity possible with dual-frequency allows a navigation system to detect when the navigation signal includes errors outside of a given confidence range and take appropriate action. In most cases, this simply means excluding any faulty data from the position calculation. Dual-frequency also allows for greater positional accuracy and, crucially, is much quicker to resolve “integer ambiguities” when it has to re-initialize because the signal is interrupted. This means that if a car goes through a tunnel or loses sight of satellites in an “urban canyon” it can reestablish an accurate position fix in a matter of seconds once the satellites are back in view. Since highly urban areas will be the primary market for autonomous vehicles, this is a serious advantage. New augmented single-frequency receivers --more advanced versions of what you have in your car today-- can match the positioning accuracy of dual-frequency receivers under certain conditions, but dual-frequency options are superior because of their greater integrity and the fact that they are not dependent on augmentation from a local network of ground-based reference stations. As the 2017 GSA GNSS Report says, “it is clear that autonomous driving technology requires highly accurate position and navigation in all scenarios. This means 100% position availability at decimetre level or less, anywhere, anytime and under any condition…” The report points out that this is only achievable through the integration of a variety of sensor types, but given the stringency of this requirement, the only reason for a manufacturer to prefer a single-frequency receiver would be the reduced cost.
Single-frequency receivers may be cheaper for autonomous vehicle manufacturers in the short term, but skimping on safety-critical technology could be a serious risk. A 2016 Mobile World Live survey found that “only 5.5% of respondents felt the vehicle owner was liable for connected car security, suggesting that security will continue to be the responsibility of vendors and service providers rather than the users themselves.” Every precaution will have to be taken to protect vehicle manufacturers and operators from liability and to protect user confidence in the safety of autonomous transportation services. Investors and adopters alike will be very sensitive to these concerns, and it will be well worth the additional investment in the safest guidance systems. GNSS will be a part of that investment, integrated into sensor packages including lidar, radar, and cameras.
The increased accuracy and integrity provided by precision GNSS will have important financial implications beyond consumer confidence and injury liability, as insurance and ride fees move to usage-based models. Many other location-based services, including tolling, weather updates, traffic information, parking availability, and stolen vehicle recovery, also stand to benefit.
It must also be noted that autonomous and connected cars will provide a massive new target for cybercriminals to steal, hijack or crash by spoofing guidance signals. Dual-frequency signals are harder to spoof than single-frequency and also more resistant to jamming and other interference. This is due in part to the increased complexity of simulating multiple signals and partly due to simple redundancy. Starting in 2018, the new European Galileo constellation will also help guard against spoofing by broadcasting an authentication signal on a separate frequency.
A 2016 market report from ABI Research entitled “Low-Cost Precision GNSS Receivers” estimates that dual-frequency receivers will be available for $50 per unit by 2021. This is consistent with the recent trends in price reduction and mirrors the autonomous vehicle deployment timelines put forward by Uber and Ford. It is also consistent with the forecast of the GSA, which said in 2015 that it expected “low-cost, multi-frequency chipset/receivers to appear on the market in the next few years…” driven by “automotive and other machine-to-machine applications.” Dual-frequency surveying units traditionally cost upwards of $5,000, but the new, more affordable receivers now available from companies like SwiftNav cost about $600. Another roughly 10x decrease to $50 seems plausible as dual-frequency receivers change from specialist, low-volume products to mass-produced car components.
There is no reason to believe that the price reductions will stop there. Many of the top chipset manufacturers are investing in dual-frequency systems for both mobile devices and wearables. Their biggest clients, cell phone manufacturers, are not interested in adding $50 to the cost of a phone for improved location services. One startup in Poland, ChipCraft, is developing a multi-constellation, dual-frequency system-on-a-chip with an integrated antenna that they project will cost only $10 per unit. That project is a long way from completion, but it is a clear indication of the direction that receiver development is heading. As ABI principal analyst Patrick Connolly has said, “As the receivers’ unit price drops below $50, we expect to see a market develop for location technology services, such as artificial reality and head-up displays in higher-end vehicles.” As the entry price for the adoption of the technology is lowered, more and more developers and users can be expected to embrace it for an increasing number of applications, including crowdsourced surveying.
Implications for Surveying
In the next five to ten years there will be massive new scaling opportunities for organizations who crowdsource surveying. Together with already available open-source post-processing software, $50 dual-frequency receivers would allow parcels to be surveyed with decimeter accuracy--even without any supporting augmentation infrastructure--for a fraction of the current equipment cost. Where satellite-based augmentation or geodetic control points are available, accuracy down to the one-centimeter level will be possible.
The rapid initialization and signal acquisition of dual-frequency--which allow parcels to be surveyed more quickly--may be an even more important long-term advantage than reduced equipment costs, which are amortized over long periods when surveying at scale. The speed of surveying can be further enhanced with the use of unmanned aerial vehicles. Drones equipped with cameras and dual-frequency receivers can rapidly create maps of land parcels. The V-Map system, from Micro Aerial Projects, LLC, uses a dual-frequency receiver to record the drone’s location each time the onboard camera takes a picture. Linking 2D images to precise GPS coordinates enables the system to use technique called Structure from Motion to create a detailed 3D map of the surveyed area without the use of Ground Control Points. In March 2017, V-Map receivers were certified to meet the international accuracy standards for first order geodetic control positioning by the National Mapping and Resource Information Authority of the Philippines. (Full disclosure: Walter Volkmann, the President of Micro Aerial Projects, LLC, served as a technical advisor for this article).
Crowdsourced surveying is already being conducted in some areas. Through funding from the United States Agency for International Development, citizens of rural African nations are using the Mobile Application to Secure Tenure (MAST) smartphone app to record formal land holdings. MAST was launched in Tanzania in 2015, expanded to Zambia in 2016, and is being rolled out in Burkina Faso. MAST uses GPS in existing cellphones that is only accurate to a few meters, supplementing this GPS data with satellite imagery. Once dual-frequency receivers are widely affordable, programs like MAST will become feasible in countries, like Jamaica, where there is a legal requirement for higher accuracy, as well as in urban and peri-urban settings where greater property density and property value demand increased accuracy and resistance to interference.
Even as hardware costs are reduced, challenges to property rights formalization will remain. It will still be resource intensive and regulations must evolve to allow for these technologies. Governments must be convinced of the trustworthiness of crowdsourced survey data. In most jurisdictions, laws require that surveying is conducted only by licensed surveyors. Crowdsourced data cannot be used to register land no matter how accurate it is. We believe, however, that as the quality of crowdsourced data converges with professional survey standards these policies will change despite resistance from entrenched interests. When trust can be placed in the integrity of the data rather than that of the operator, the argument for restricting surveying services to licensed professionals will be weakened significantly.
More accurate surveys will also help avoid problems down the line when the formalized land is bought, sold, or inherited. When conflicts arise between owners who were not party to the original, mutually-recognized, boundaries, inaccurately surveyed land can lead to costly disputes. In such cases, there would be a real risk of deformalization. If a land title is less secure it is less valuable, and there is less incentive to keep an insecure parcel registered.
As we noted in our last article, there aren’t enough professional surveyors in developing countries to help formalize all the land that needs attention. Only crowdsourced surveying, enabled by tools like dual-frequency receivers and programs like MAST, can bring about significant progress. The potentially transformative impact of these technologies is hard to overstate, and increasingly hard to overlook. The GSA’s May 2017 summary of emerging trends in the GNSS surveying market begins simply: “Falling device prices drive the democratization of mapping.”