During a test drive near Ford's headquarters in Michigan, the test team noticed a strange phenomenon in their unmanned vehicles. Every one of the unmanned cars is slightly turning at the same point on the lane, as if it were avoiding pits, and the study found Sensor that the problem was not in the car - but in the map. The test team has just updated the 3D map of the test route to guide the unmanned vehicle. But a small fault causes the data on one pixel on the map to be incorrect. The map tells the car that a point on the ground raised 10 inches, and in fact the ground is in a perfectly horizontal state. The test team used a new map, the naked eye looks perfect, but for the eyes of unmanned vehicles is not the case. A simple error pixel is enough to cause the car to break down.
Internet companies and car companies are investing in a new generation of maps, which are often referred to as HD maps, far more than the basic segmented navigation. They will continue to update the lane markings, street signs, traffic signals, pits, and even roadside height data together - all of these data are accurate to centimeters. In order to meet the higher requirements of the user - unmanned vehicles. This new map puts the car in a more accurate world and greatly enhances the sensor's ability to sense. For example, Google Maps can remember your parking location. When the lane mark is covered by snow, or when the truck blocks Fuel Rail Pressure Sensor the traffic signal of the car, the map can guide the car. But also let the car sensor empty out, focusing on the map does not include the target, such as pedestrians, animals. The map is not as impressive as the unmanned car video on the road, or like the center of the major litigation problem with Google (Google, Tech30), the LIDAR laser sensor is eye-catching, but the map is under the car's hood A key element, especially in the autopilot era. Over the next few years, the map will help ensure a more secure deployment of unmanned vehicles, at least to create a new billions of dollars.
Google, Optics, Ford and others have used sensors in their unmanned vehicles to collect data for making HD maps. If you think unmanned cars will eventually occupy the world, then you need to every city and every street for a very accurate drawing. The current high-precision map of the status quo is, according to the description of excellent website, excellent unmanned team is currently only a high-precision mapping of a city: Toronto. With the intensification Pressure Sensor of competition, there are still questions about how these high-cost map prospects are still in the industry. Some people think that with the depth of the depth of learning technology, the car will eventually become smart enough, do not need to rely on a wide range of maps.
If you have a car, then the car navigation map must be necessary. Navigation map will become a core component of the car, and will continue to produce economic benefits. If you believe that the autopilot can eventually travel anywhere, every city and every street will need to be planned in detail.