5G has a prominent role to play in positioning for automotive applications. In particular in dense urban canyons, it could complement GNSS-based positioning, enhancing positioning performance by leveraging the expected dense base station networks to deliver highly accurate location services. By analyzing the time difference of arrival of cellular signals from multiple base stations, receivers will, for example, be able to determine their position relative to them using multilateration, while advanced approaches using antenna arrays could be used to implement angle-of-arrival and angle-of-departure based positioning solutions.
Real-time high-definition maps
Cameras and lidars are heavily relied upon by autonomous vehicles to perceive their environment, detect obstacles, and inform decisions on the safest way to reach the intended destination. In order to determine their absolute position in space, the vehicles’ autonomous driving computers can match up the data they sense with a high-definition (HD) map of their surroundings, which offers a detailed representation of the surrounding environment. SD maps, offering meter-level resolution, have long been used by OEMs to enable features such as speed assist and predictive powertrain control of trucks to reduce fuel consumption by anticipating upcoming curves and slopes on the road and other functions related to automatic cruise control. HD maps, which are required to implement advanced ADAS Level 2 and Level 3+ features, demand decimeter-level accuracy of the map, as well as of the position of the vehicle and surrounding objects. To ensure that the data can be used by the autonomous driving systems of as many vehicles as possible, HD maps are typically designed to comply with the ADASIS v3 interface specification, which is the de-facto industry standard used by OEMs around the world.
As the central repository of static geographical information surrounding the vehicle, the trend is for HD maps to increasingly take advantage of low-latency and reliable data links to deliver dynamic information as well. To this end, they will have to be continuously updated, crowdsourcing data from as many vehicles’ sensors as possible, and making it available to vehicles around them. To do so, vehicles will be able to publish relevant information from their sensors to the base map, which is made available on the cloud.
Accurately positioning obstacles detected by vehicles on the HD map is essential, yet challenging, in particular when traveling at high speed down a highway lacking obvious landmarks. High precision satellite-based positioning becomes a necessity to first accurately and reliably locate the vehicle itself, determine the relative position of other obstacles, and warn nearby vehicles.
No single sensing technology can come close to matching the performance of a human driver. Human brains intuitively combine a broad variety of inputs to draw meaningful conclusions. Machines need to be instructed precisely how to fuse the data their sensors gather, leveraging expertise more commonly available in data-heavy industries than in the traditional automotive sector. But when well done, sensor fusion allows to increase the reliability of sensed data and extract information that only becomes available when multiple sensors are used in conjunction.
Positioning is one area that can benefit from sensor fusion. Accurate satellite-based positioning requires constant line of sight between the vehicle and the orbiting GNSS satellites. Underpasses, tunnels, dense forests, mountainous terrain, and urban canyons can all degrade positioning performance, ultimately leading to signal loss in indoor environments. By augmenting satellite-based positioning with inertial sensor measurements used to reconstruct the vehicle’s trajectory in dead reckoning solutions, intermittent gaps in GNSS coverage can be bridged.
This approach can be extended further by incorporating camera, lidar, and radar data that allow the vehicle to position itself relative to landmarks, as well as using the HD map, and potentially indoor positioning solutions based on ultra-wideband, short-range, or cellular communication technology. Together, these complementary solutions can extend the availability and coverage of reliable and precise positioning to a broad variety of environments.
By taking advantage of redundancies, sensor fusion can also help stabilize the vehicle’s overall sensing performance. Sensors can drift, go offline, or fail altogether. By constantly cross-checking the input across sets of sensors, machine learning algorithms can be trained to detect incongruities and flag sensors that cannot be relied on. While the situation may be easily remedied by recalibrating the sensor in question on the fly, it could also signal that the sensors are deliberately being jammed or spoofed by hackers.
Augmented and virtual reality
Augmented reality is transforming in-vehicle infotainment systems by projecting relevant information right into the driver’s field of view. Mercedes Benz, for instance, already offers an advanced AR heads-up display in some of its latest vehicles, providing information on the speed limit, the driving speed, and GPS navigation instructions Panasonic has presented its latest heads-up display solution, offering crisp 4K projection that follows the gaze, delivering even more content, including advanced collision detection to increase traffic safety, visual highlights of relevant environmental information, and easy-to-follow navigation instructions to help keep the driver’s eyes focused on the road.
Unlike augmented reality, which superimposes artificially generated information into a user’s field of view, virtual reality fully immerses users into a virtual world. Innovative companies, such as holoride, have found creative ways to leverage high data rates and low latencies to build next-generation passenger entertainment solutions: Using a headset, passengers can experience their journey as a first-person adventure game in which the hero’s movements are perfectly synched up with the car’s motion. We often forget that our smartphones can make phone calls. If holoride’s immersive experiences catch on, we might forget that our cars were originally designed to get us places.