As we move further into the era of autonomous vehicles, the market opportunities for GNSS receiver developers are multiplying.
Every self-directing vehicle needs to know its position: from ships and drones to self-driving cars and delivery robots. The best way to obtain an accurate, absolute position on the earth’s surface is by receiving signals from one or more global navigation satellite systems, so a highly accurate GNSS receiver is an essential feature of every autonomous vehicle.
Manufacturers are developing advanced solutions, introducing multi-constellation, multi-frequency receivers with unprecedented levels of accuracy, robustness, integrity and precision. Nothing less is acceptable in a market where safety of life is a critical consideration.
GNSS must work in concert with other sensors
But GNSS alone isn’t enough to keep an autonomous vehicle navigating safely and reliably. It also requires inputs from an array of on-board sensors: radar, lidar, cameras, inertial sensors (gyroscopes and accelerometers), and also signals from nearby Wi-Fi points and cellular base stations.
The inputs the vehicle receives from GNSS will need to be fused with the inputs from these other sensors and signals to compute an accurate position (precision) and to check for consistency among these sources of navigation data (reliability). This complex sensor fusion means that GNSS receiver developers who are serious about the autonomous vehicles market have new test challenges on their hands.
Developers can no longer test receivers in isolation
There was a time when receiver developers could focus purely on the performance of the GNSS receiver itself. Testing was a matter of ensuring the receiver met certain criteria: time to first fix; static and dynamic position accuracy, signal acquisition sensitivity, and so on.
All of these criteria are still fundamentally important – the receiver must be able to acquire and process GNSS signals in the way the customer expects, and receiver manufacturers must be able to produce documented evidence of their products’ reliability in these key areas.
But in the world of autonomous vehicles, the GNSS receiver is just one subsystem in a complex system of sensors, actuators, software applications and algorithms. Any system integrator will want to be 100% certain the receiver can function effectively within this wider system which fuses the GNSS data with data from other sensors and signals.
Developers who want to get ahead in this market, therefore, should work with integrators to test how their receivers perform as part of a wider vehicle self-directing system.
Two key aspects of sensor fusion to test
There are two key aspects of sensor fusion to test.
The first is the ability of the system to combine all of the sensor inputs to compute a position more accurately and more reliably than any single sensor or sub-combination of sensors could achieve on its own. This is by no means a given: all sensors and signals can produce noise and error, and in a poorly-configured system, these error states can amplify each other rather than compensate for each other – a deleterious condition known as adverse uncertainty propagation.
The other key aspect to test is the handoff from GNSS to other sensors when the GNSS signal is unavailable – such as in a tunnel or indoor car park – or compromised by a threat source. There are many threats to GNSS, and the signal can be disrupted by anything from tall buildings to deliberate jamming and spoofing. The receiver must recognise when the signal has been disrupted or tampered with, stop trying to acquire it, and start acquiring it again when it’s safe to do so. Developers that already include onboard MEMS sensors in their receivers will be familiar with basic sensor fusion testing, but the key now will be to test handoff to (and back from) other sensors outside of the receiver module.
Co-simulation is essential – and Spirent and its partners are leading the way
To evaluate sensor fusion effectively, the lab setup must be capable of co-simulation that is simulating all of the different signals, the sensor inputs and the driving environment together, in real time. As well as the normal environment, co-simulation should be able to simulate many different types of interference and disruption, so that handoff and back can be tested in a wide array of realistic scenarios.
With many different pieces of test equipment in the mix, co-simulation is highly complex to do.
Spirent has been leading the way, integrating our multi-constellation, multi-frequency, low-latency GNSS signal simulators with real-time systems, such as dSPACE SCALEXIO, automotive simulators, such as dSPACE ASM, IPG CarMaker, Oktal SCANeR, Mechanical Simulation CarSim, and realistic environments such as Sim3D, to create a co-simulation environment for autonomous vehicle hardware-in-the-loop testing.
Our test orchestration, scenario generation, analytics and visualisation software give developers fine-grained control over the test environment and visibility of test results, while our test automation solution, TestBench, enables developers to accelerate time to market.
Spirent continues to work with industry to develop powerful, flexible and future-proof solutions for sensor fusion testing. Our SimINERTIAL software already offers the ability to simulate the output of inertial sensors coherently with GNSS, and we are working with Warwick University, Chronos Technology and WMG to integrate Wi-Fi signal simulation into our solution.
At the same time, we continue to ensure our GSS7000 and GSS9000 simulators offer all of the latest GNSS signals and ICDs (including continued provision of the Galileo PRS signal after Brexit), so developers can always be assured of testing with a fully-featured and realistic RF environment.
Questions about sensor fusion testing for autonomous vehicles? Get in touch.
If you’d like to learn more about how Spirent enables sensor fusion testing for autonomous vehicles, or if you have any other questions about testing with Spirent, please do get in touch.