The Mapping Problem for Autonomous Cars


Why do we need Autonomous car Maps?


In everyday driving, there comes a point when sensors like radars or cameras stop seeing due to any hindrance ahead that`s when the inbuilt maps can convey helpful data. For instance, a camera is not programmed to see a speed limiter sign hidden behind a truck but highly detailed (HD) maps can provide that information. The newly developed cutting edge technology sensors may convey dependable data up to 300 meters or even more, the maps could allow the car to see even farther than that.


A few difficulties must be overcome to get this going, a significant number of which have nothing to do with advancements, however more to the complete self-driving car ecosystem. One automaker alone might not have enough autos making a course for accumulate the information with the normal level of unwavering quality. Thus, automakers are collaborating to tackle this issue together. The combined acquisition of Nokia`s Here by VW, BMW and AUDI in June 2016 is a testimony to that. The automakers have understood the need to fabricate an effective car-to-cloud-to-car circle, where they ought to convey sensor information standardized to an autonomous gathering who could collate the data, analyse it, and transmit it back to their respective servers.


The data required for these HD maps is significantly wealthier and assorted than what is utilized for basic GPS route. Both the spatial exactness and the required recurrence for refreshes are still in face off regarding, yet they doubtlessly are unique in relation to what is accessible in the present.

The list of problems


  • Issues with data transmission- A self-driving car collects more than one terabyte of data a day. With that highly detailed information coming from the numerous car sensors; it is not cheap to send it via a network like the Internet. The cumbersome data storage is just one of the technical problems that many of the brightest engineering minds are occupying. Many companies have not discovered how to actually store their data, which is why autonomous vehicles are geo-fenced. Physically they cannot adapt to the data in the trunk of the car, so they are limited to certain areas.


  • No common standards: There is no common standard for these top quality 3D maps, nor does any data exchange since organisations consider maps as important property information. Everyone is trying to develop their own internal HD mapping solution to meet their autonomous driving needs, and this does not stop, everything is reinventing the wheel, and this is wasting a lot of resources


  • Strict laws regarding data collection in different region: Although digital maps are widely available today, there are countries around the world that largely regulate their geospatial data and even prohibit the export of geospatial data. This problem is inherent to normal navigation maps. Although most of these maps are not available directly to consumers, they are designed to communicate with the robot’s brain, not with the passenger; the high level of detail it contains could generate alarm among privacy advocates. Mapping and information exchange between vehicles are two areas in which the government will be more involved.


  • Regular updation of maps: The major challenge for self-driving car maps is to keep them updated continuously, so that they provide the latest information to the cars. Unlike conventional digital maps, self-guided maps require almost constant updates. The slightest variation on the road, a construction zone that opens during the night or a bit of debris could stop a car without a driver. For full autonomous cars to be deployed, it needs to have a high-definition map of the area, like a map annotated with what are the permanent fixed objects in that area.


  • Slow response to climatic variation: The photonic and imaging solutions cannot yet be trusted to perform the recognition and scope required for self-driving vehicles independently and under all weather and lighting conditions. The calculation of distance, or interval, remains a challenge for single-chamber methods, even if stereoscopic systems are improving, they still have a range limit. As a natural consequence, this means that mapping and GPS systems depend to a large extent. But with the very low signal strength of satellite GPS signals, storm dropouts, wooded or very mountainous areas that have reliable data from both maps and artificial vision are essential


  • High cost of HD mapping: The companies at present develop maps by driving cars furnished with LIDAR rotating units mounted on the top of their cars that shoot lasers, making pictures of the street and nature. Engineers, in a time-consuming process, review images and label objects that are found, such as stop signs, buildings, traffic lights, and non-entry signs. The laser equipment needed to perform this scan is expensive; it can cost a lot to equip on only one car to do this job.


The information has been sourced from our report titled Global Self-Driving Car Maps Market 2018-2023.  Download free sample to know more

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top