SAN FRANCISCO: California-based start-up Artisense has developed software for vehicles and robots that can create and update maps used for automated driving. The project is currently being piloted in Berlin, and the company hopes to set an industry standard with its product.
Artisense's algorithms create 3-D maps in real time while the car is navigating through a city. But what really makes this software different is that it doesn't need expensive hardware and only uses a fraction of the power that similar programs require.
"We want to help rethink mobility and put ourselves at the peak of development," says Till Kaestner, one of the company's founders. He says that the goal is to help machines learn how to see like humans.
Other concepts in autonomous driving use photorealistic imagery to navigate, which ends up being one large picture file. Artisense creates a map of the streets by using "cloud-based geospatial intelligence." This means the map relies on artificial intelligence.
The onboard computer in most cars is capable of using the Artisense software, which doesn't even need a GPS signal or expensive sensors.
Finding a way for vehicles to navigate a busy city without help from a driver is a quest many developers have been on for a while.
Companies that create navigation systems, such as Here, TomTom or Google, use large amounts of data to create maps of the world, which are made with the help of laser radar systems, sensors and vehicles with cameras driving through cities. These methods create high-quality, precise maps that can be updated with real-time data as needed. This allows for autonomous cars to change routes if there's bad traffic or construction work going on along the designated route.
The problem with these methods is that the images create huge data amounts, which means cars need certain chips and algorithms installed from the beginning that can process this large amount of data.
This is where Artisense is different: The recorded data can be saved on basic hardware that can be integrated into pretty much any car - and according to Kaestner, it's much cheaper than the others as well.
Artisense has found some of its first partners to trial the software via the innovation platform Beyond1435. Some of the partners include Siemens, aerospace firm Bombardier and recycling company Alba.
Bombardier wants to use the software to develop safer train systems that can better recognize hazards and obstacles, as well as for machine-learning technology. For Alba, the system should help create safer roads, which it says is key to the future of driverless cars.
Artisense's headquarters are in Palo Alto, California, but its origins lie in Munich, as the company was originally founded as part of the Technical University of Munich.
Vehicles with Artisense technology have been roaming around Berlin since January and should be done creating the maps by March. "We are very happy with the way the pilot project is going," says Kaestner. "Every day, we cover around 400km on the streets of Berlin."
"Now it's about closing all the gaps in the maps of Berlin and then starting to work on concrete-use cases based on the sensor data."