The role of GNSS in the age of autonomous driving

This article originally appeared in the Feb'23 magazine issue of Electronic Specifier Design – see ES's Magazine Archives for more featured publications.

When was the last time you got truly lost while driving your car? It’s probably been a while, at least if your car comes with an in-vehicle navigation solution, or if, instead, you use the one in your smartphone. Not getting lost is just one of the many ways global navigation satellite systems (GNSS) technology has affected, and improved, our daily lives. And there will be many more, particularly as we enter the age of autonomous driving.

The method GNSS receivers use to precisely pinpoint their position on the planet has hardly changed since the 1990s when the technology was first conceived. First, they calculate the time it takes radio signals emitted by at least four dedicated GNSS satellites to travel to them from orbit at the speed of light. Then, knowing the exact location of each satellite the instant its signals were emitted, the receivers can use the travel times to determine their own coordinates.

But while the underlying approach has remained the same, the GNSS satellites, the receivers, and the supporting infrastructure, services, and algorithms have all improved dramatically over the past three decades. In the 1990s, GNSS hardware was bulky and slow, and primarily used by the military. Fast-forward to today, and the ubiquitously used hardware is so small that a handful of high performance GNSS receivers can fit on a single fingernail.

This article will take stock of how GNSS technology has matured as we enter the age of autonomous driving. It explores how these technological trends are transforming vehicle navigation solutions and how they are contributing to making autonomous vehicles a reality. And finally, it looks at how GNSS-enabled autonomous vehicles are already impacting businesses.

GNSS trends in vehicle navigation

GNSS technology first entered our vehicles via satellite navigation, and today, almost all new cars come with the option of an integrated navigation solution. But while they are well established, today’s navigation solutions still have room for improvement, as any avid user of the technology will confirm. One common issue: we often need to wait too long for the GNSS receiver to determine its position and for the navigation solution to provide the quickest route to the destination when we turn the ignition.

Responsiveness is one challenge. Accuracy is another. Today’s navigation solutions can reliably locate vehicles on the right road. Tomorrow’s will be expected to locate them on the right lane, a prerequisite for more accurate navigation guidance and new use cases such as augmented reality heads up display (AR-HUDs). AR-HUDs project relevant information including navigation guidance, collision warnings, traffic status, and more right into the driver’s field of view.

These twin challenges are being addressed with technological solutions. To increase their responsiveness, GNSS receivers can, for instance, download satellite data required for positioning via the high-bandwidth cellular network using A-GNSS (assisted GNSS) rather than having the slow satellite download delay operation by half a minute or more. Additionally, storing the vehicle’s last position and retrieving it when the system is initialised can speed up performance and improve the user’s experience.

Meanwhile, the adoption of GNSS receivers capable of concurrently tracking multiple global GNSS constellations –the United States’ GPS, Russia’s GLONASS, Europe’s Galileo, and China’s BeiDou – has increased the number of satellites receivers can see, even from deep within urban canyons. Tracking them on multiple frequency bands brings additional benefits, helping to mitigate satellite signal delays incurred in the charged troposphere, and reducing the impact of GNSS signals that bounce off buildings on their way to the receiver – two of the most important sources of positioning errors.

But even multi-constellation, multi-band GNSS receivers are unable to continuously provide the lane-level positioning accuracy required by AR-HUDs. In order to robustly achieve accuracies in the tens of centimetres, GNSS receivers need real-time GNSS correction data to correct satellite clock and orbit errors and other error sources. The precise-point positioning real-time kinematic (PPP-RTK) correction services and the algorithms GNSS receivers use to process them are now finding their way into cars that will hit the roads in the near future.

Then, they need to be able to deal with interruptions in service availability, for instance when driving through a tunnel. Dead reckoning solutions, which feed data from inertial sensors – accelerometers and gyroscopes – and vehicle sensors to the GNSS receiver, can bridge short interruptions in coverage.

Finally, the navigation solutions need a way to gauge the quality of the data they receive from the GNSS receiver. Some new receivers offer an additional output called the protection level, a concept borrowed from the aviation industry, which gives a highly reliable upper limit on the maximum position error. This lets AR-HUDs and navigation systems postpone guidance if the GNSS signal quality is temporarily poor.

GNSS trends in autonomous driving

If not getting lost on the way to a holiday was nice, not having to hold onto the steering wheel will be a blessing. Although we are still many years from the commercial availability of fully autonomous vehicles, advanced driver assistance systems (ADAS), which support drivers with a growing range of features while still expecting them to remain in control of their vehicle, are now commonplace in the line-ups of most major automakers. But the transition from advanced ADAS features to the actual eyes-off-the-road, hands-off-the-wheel autonomous driving is proving to be trickier than anticipated.

More major automakers are recognising that GNSS technology has an important role to play in supporting that transition. Before fully autonomous driving features roll out everywhere, they will most likely become available on dedicated stretches of the road network and under very specific operational conditions. Take the highway chauffeur, one example of an autonomous driving feature, which takes control of the vehicle on precisely geofenced sections of the highway.

Determining whether the conditions are met to enable the functionality is a critical decision, as you wouldn’t want your highway chauffeur to speed up your vehicle when you are not on the highway itself, but on a country road right next to it. The consequences of falsely identifying the so-called operational design domain of an autonomous driving functionality are imposing ever more demanding requirements on the GNSS receiver than the lane-level navigation use case described earlier.

The next trend in GNSS technology will likely involve developing safe positioning solutions. These solutions have to comply with industry standards for functional safety to ensure that all the ways in which a vehicle’s hardware and software might fail is understood in detail, and that these failures can be detected and mitigated. Additionally, the weaknesses inherent in the technologies themselves need to be well understood, such as the impact of reflections, weak signals, or misuse by their users. And finally, the solutions will have to be resilient against cyberattacks.

As autonomous vehicles become more capable of safely navigating through more and more challenging scenarios, the operational design domain of each functionality they offer will grow until, ultimately, they overlap and cover the entire road network in all environmental conditions. The result: full autonomous driving, all the time, everywhere. How soon this scenario will become reality, assuming that it ever will, is anyone’s guess. But even before it does, autonomous driving will impact how businesses operate. In fact, it already has.

The benefits that autonomous vehicles bring businesses are clear: reduced personnel and operational costs, increased safety, and overall optimised efficiency. Fleet owners, for instance, will be among the best placed to offset the initially high price tag and per-mile cost of autonomous vehicles by maximising the time the vehicles spend on the road while saving insurance fees, salaries for highly qualified drivers, or by eliminating the drivers altogether, potentially replacing them with remote supervisors. A case in point: remotely supervised robo-taxis have already made a name for themselves in several cities around the world.

The trucking industry will likely also become an early adopter of the technology, with many of the big names in the business actively developing autonomous vehicles. While it may take years until autonomous long-haul trucks transport goods across continents, several trials have demonstrated the feasibility of the technology in harbours, motorways, and other public roads. Additionally, platooning, where several trucks automatically trail each other in a convoy has demonstrated its potential to reduce drag, save fuel, and cut CO2 emissions.

A growing role with increasing automation

Only three decades in, GNSS technology already has a storied history to look back on. First conceived as a military application, the technology was quickly adopted by the automotive and transportation sectors and specialised industries such as surveying and precision agriculture. It later found its way into our smartphones, smart watches, and a plethora of consumer and industrial IoT devices. And after initially helping us chart the fastest course to our destinations in our cars, it will, at least eventually, contribute to getting us there autonomously.

As more features in our cars depend on high precision and safe positioning, the GNSS receiver and all of the related location services will only become more important.