Unmanned vehicles will require qualitatively different cards

Nokia HERE shows the result of 3D road scanning

Unmanned vehicles need fundamentally different cards than those that people can use. First of all, we are talking about resolution: if the error in a few meters from the GPS for a car navigator does not play a big role, then for robotirugs, information with an accuracy of centimeters is required. These maps should be updated as often as possible, ideally in real time with information about incidents and road closures.

Sergey Brin claims that fully autonomous cars will appear on public roads in 2017, more conservative forecasts speak about the 20s of this century.

Already today, adaptive cruise control and automatic parking systems are becoming more and more independent. The future car will not require any human involvement - you can read a book while driving, if, of course, you manage to relax in confidence in driving a machine. The appearance of such machines is associated not only with technical difficulties, but also with legal ones: if an accident happens, then who is to blame? Another important element is maps.


HERE data collected from cars on the Berlin Highway
')
HERE is a cartographic division of the Finnish company Nokia. Map data is collected using satellite and aerial imagery. Each month, 100 billion anonymous coordinates of points are processed from satellite-navigation devices from partner companies. This allows you to determine the speed and direction of movement.

But the most detailed data is collected using hundreds of machines with cameras and lidars, laser devices with rangefinders. Coordinated this fleet of cars from a small building on the campus of the University of California at Berkeley.

While driving, the cylinder on the roof rotates, fires 32 laser beams and analyzes the reflection of light. In a minute going to 700 thousand points. The inertial system records the position of the vehicle, and the data from the lidar are corrected for these parameters. The resolution of the device - a few centimeters.

The markings and road signs stand out well due to the use of reflective paint. To isolate characters from the resulting amount of data, a computer vision system and people’s manual labor are combined.

In total, HERE has about 200 machines that John Ristevski has created, there are still a similar number of machines with an older version of equipment. Ristevski showed Wired a car. This is a brightly colored Volkswagen Jetta with a variety of sensors on the roof. At the foot of the passenger in the front seat is a battery and a Linux computer with a terabyte hard drive.

HERE cars have traveled 2 million kilometers in 30 countries on 6 continents - and this is only in 15 months. Google, HERE’s main competitor, is also working in this direction, but the search giant primarily scans areas close to the company's offices in Mountain View.

Here is a demonstration of the result of Nokia’s efforts. This is a three-dimensional model of New Orleans.



High-resolution maps are needed to help understand the robot car that expects it on the way. If you rely only on information from sensors without any existing data, then the system will be more difficult to work.

Road conditions are changing rapidly, and another challenge for map services is the rapid reflection of accidents and road closures in their products. The ideal time is as close to real as possible. Future autonomous cars may report obstacles on the road to each other, although problems may arise with response times from the server. In theory, a swarm of unmanned vehicles should be able to quickly change its route.

One of the main problems with “smart” cars is the fear of new technology among ordinary users. According to the survey , 88% of Americans would be anxious while in a moving robotic car.

But if the car warns that it will now change the lane in order to avoid an accident ahead, and then actually do it, the passenger will quickly develop confidence in the vehicle. HERE lead designer Peter Skillman showed a couple of videos to demonstrate this principle. On the first one before the change of the strip a warning arrow appeared about it. In another video in the style of the atlas of highways, a trip to Chicago was shown, during which passengers could see where they were and better understand why this or that action happened.

Based on Wired .