About Cloud to Car mapping
It is obvious that GPS-based 2-D maps are not accurate enough for new generation cars that do not have drivers. When dealing with a regular map, the position of an object can be accurately identified till about few meters. HD maps ensure positioning up to 10 centimetres. The need for precise positioning and positioning requires real-time or near real-time updating of the mapping environment, which is possible only through the introduction of mapping systems between the cloud and the car.
The maps for the driverless cars must be very detailed (HD), containing all the critical characteristics of the road, including the slope and the curvature, the types of marking of the lane and the objects that are on the edge of the road. Moreover, it is necessary to build the mapping system for autonomous cars that will become part of the driverless car software, to ensure that the vehicle can run independently, and at the same time allowing the rich and almost real-time incorporation of contextual awareness of traffic situation around the vehicle.
The mapping systems between the cloud and the car allow manufacturers, suppliers and users of driverless cars to obtain the latest information on the road network to locate position and manoeuvre the vehicle automatically. This new information is provided via “air” software updates. Live HD maps that are constantly updated in the cloud allow highly or fully automated vehicles to choose optimal driving strategies based on road profile, lane curvature, terrain, congestion, detours and other factors that impact movement of vehicle. To shape detailed environmental models, mapping experts must develop Artificial Intelligence for cloud mapping, as well as process huge sets of data and ensure real-time data communication between the cloud and automobiles.
Who is doing this type of mapping?
Many mapping companies are following this method. For example, Baidu has incorporated a data collector for HD maps in its latest iteration of Apollo 2.5 (Open source platform for autonomous cars) launched in April 2018. Here, on the other hand launched HD Live Map in 2016 which stores data coming from car`s sensors and cloud so that a car can get the best from both. The reason behind this approach is, although the sensors can feed real time data but they don’t have any memory so when you start feeding the real time data in the cloud, the maps could be updated immediately.
More than 10 companies have created highly detailed, multi-layered road and highway maps, and there’s still a lot of work to be done. As a general rule, an HD map is created with the help of several cameras and sensors installed in the machines that collect data. The collected data is then analysed in real time with powerful integrated computers capable of processing the input of multiple sensors. Thus, data is compressed and transmitted to the cloud. The cloud server totals and accommodates transmissions of transmitted information and produces the cutting edge HD maps accessible in the cloud, important to make a mapping system for driver-less vehicles. Any HD map has several levels, which contain static and dynamic information, which form the map of the vehicle’s environment.
Some companies are using a mixed approach for the mapping exercise. For example Qualcomm partnered with TomTom in Q1 2017 to facilitate crowdsourcing approach to act as a source for map generation. It is collecting data from thousands of connected cars which are already on road today and simultaneously building a cloud based platform to store and update the maps. But, the matter of the fact is that even this alternative approach can’t do without cloud. Thus, we believe that cloud to car mapping is the way to go for HD mapping.
The information has been sourced from our report titled Global Self-Driving Car Maps Market 2018-2023