How ‘swarm' data improves vehicle mapping and safety

RENNINGEN, Germany — Connected map services are helping automakers enrich in-car mapping using sensors and "swarm" data — information crowdsourced from many vehicles on the road.

The data is transferred into the clouds of software-defined vehicles to provide real-time driving information and increase safety and driveability.

"Connected maps are an extremely important part of the swarm data concept," Pedro Pacheco, vice president of research at Gartner, said in an interview. "This kind of concept is definitely the future," he said, and "will bring major benefits" for automakers.

It starts with swarm data collection. The data extracts insights, which go to the cloud to enhance vehicles' intelligence, including the performance of advanced driver-assist, navigation and other systems.

Volkswagen is among the automakers using swarm data for advanced driver-assist systems. VW's Travel Assist system positions the vehicle in its lane and maintains speed using the swarm data that's available.

Connected maps provide the cornerstone of swarm vehicle intelligence. For example, Bosch's connected map services combine swarm data with other information from a car's infrastructure to provide a virtual sensor for driver-assist systems. The map sensor aggregates real-time data onto four map layers.

"We are generating live maps using the car's intelligence and this is really revolutionary," Thomas Classen, project leader for Bosch's Automated Driving Alliance, said at a tech day event at the company's headquarters here. "The car sees what is around it and gives a very precise localization."

The map layers include a planning area, with road line geometry and other features like traffic lights. A behavioral map layer with swarm trajectories shows a car's optimal driving speed. Key benefits include a reliable map sensor in all weather conditions, which compensates for a driver's obstructed line of sight.

Multiple automakers, which Bosch would not identify, provide the data, allowing the supplier to move closer to an open map ecosystem. Since January 2021, 2 billion kilometers' worth of data have been harvested.

Enabling cars to collect data to automatically update maps is "a powerful concept that can even lead to many more things," Pacheco said.

For example, standard navigation systems provide updates to let drivers know if there's an accident ahead. But the information often arrives too late to change course. When cars can collect data, one can capture an accident on its camera and share that insight with other connected cars on the road, giving other drivers more time to avoid that area. Likewise, if bad weather has made a road surface dangerous, a car can provide a real-time update.

While some automakers are starting with basic swarm data collection, others already have "a good grip on this," Pacheco said. They include Tesla, which is using computer vision to identify situations using a car's cameras.

"This is an area where OEMs have in general already realized there is potential," Pacheco said.