Guest article: ADAS card – The seventh sense for assistance systems

Advanced driver assistance systems (ADAS) are the reason why vehicle sensors exist. Parking aids use cameras and radar, adaptive cruise control relies on radar, intelligent speed assistance systems (ISA) need cameras to detect speed limit signs. Lane departure warning systems also rely on sensors. Sensors are essential for autonomous driving and automated driving functions, as the vehicle uses them to record its immediate surroundings. However, sensors have a limited field of vision and cannot see beyond their own horizon. The vehicle therefore needs another technology that can map the conditions beyond the sensor horizon: a digital map. It is the seventh sense in the vehicle.

ISA systems need maps: One application example is the ISA speed assistance system, which is mandatory in new vehicles in the EU. This uses cameras to identify signs that set speed limits. It becomes more difficult if the sign is overgrown or covered with snow in winter. Implicit speed limits, such as town signs, are also not recognised as such by the camera.

This is why ISA systems are equipped with digital maps. They show the position and content of signs. These are not only the round signs on which a speed limit is indicated, but also place name signs or play streets. The position of the signs is also important, especially at motorway junctions or exits where a different speed applies to the lane leading off the motorway than to the main carriageway. Only if the position of the signs is correctly marked in relation to the course of the motorway can contradictory information on the maximum speed be avoided.

‘Ground truth’ for autonomous driving functions: Semi-autonomous functions in which the vehicle accelerates or brakes independently are becoming increasingly common. The decision as to whether the vehicle executes a manoeuvre depends on sensor information. However, sensors cannot see behind a bend or over a crest. If a stop sign or roadworks are waiting there, the vehicle could reduce its speed beforehand. Such information is not available in good time via sensors alone. However, they are recorded on a map and the vehicle can plan braking manoeuvres or alert the driver. The map thus enables anticipatory driving.

As a seventh sense, the map not only looks beyond the sensor horizon. It also functions as ‘ground truth’ and verifies sensor information. It confirms whether a sign is really in the right place or whether the carriageway is as wide as the sensor claims. It can happen that two sensors provide contradictory information. The map then provides clarity and verifies the correct information.

A standardised map for the software-defined vehicle: In the software-defined vehicle, all sensors ideally use a standardised map. They provide the map with data and the assistance systems obtain their information from there. This map is a central component of the architecture in the software-defined vehicle. All systems that require location-based information access it. This includes navigation and assistance systems as well as systems that are required for autonomous driving functions. If these systems only access a single, standardised map, they do not receive conflicting information from different maps. At the same time, it is much more economical to maintain just one map and have a single source of truth instead of multiple sources of information.

Damit Änderungen aus der realen Welt schnell und zuverlässig ihren Weg in die Karte finden, ist KI nötig, um die schiere Menge an Daten zu bewältigen: So werden täglich 35 Milliarden Datenpunkte aus Fahrzeugen und Sensoren durch KI und Machine Learning validiert und aggregiert. Insbesondere für diese Überprüfung und Zusammenführung der Daten ist KI entscheidend, um aus der Datenflut eine valide und immer aktuelle Karte zu erstellen. Bei dieser Form der KI-gestützten Kartenerstellung ist HERE Technologies mit seiner UniMap-Technologie Vorreiter.

To ensure that changes from the real world find their way into the map quickly and reliably, AI is needed to cope with the sheer volume of data: every day, 35 billion data points from vehicles and sensors are validated and aggregated using AI and machine learning. AI is particularly crucial for this verification and consolidation of data in order to create a valid and always up-to-date map from the flood of data. HERE Technologies is a pioneer in this form of AI-supported map creation with its UniMap technology.

Assistance systems play a central role in software-defined vehicles. They use sensor information, but this is not enough for autonomous driving functions. The systems need an up-to-date and reliable directory of location-based information that they can all access and enrich. If a subset with relevant content for assistance systems is derived from this central map, the vehicle will have a seventh sense that enables it to look beyond the sensor horizon and ensures that assistance systems maximise their benefits.

This guest article by Sjoerd Spaargaren (Product Marketing Manager at HERE Technologies) was published in the June issue of eMove360° magazine. Download the free PDF here or order the print version at sabine.metzger@emove360.com.

[DISPLAY_ULTIMATE_SOCIAL_ICONS]