Karlsruhe-Bangalore, Oct 19’ 2020 - Autonomous driving market has created a huge disruption in its associated industries, wherein mapping & navigation is one of the potential industries behind the scene. Achieving the fully autonomous vehicle dream requires a precise self-localization and real-time navigation guide. Mapping and localization play a crucial role in guiding autonomous vehicles with complete road scenarios and foresight the near expected situations on road.
While the traditional automotive mapping and navigation companies like Google, Baidu, HERE Technologies, and TomTom are exploring the opportunities in autonomous driving technology, new entrant like Atlatec is bringing a lot to the table with a quite confident and simplistic approach of creating high fidelity 3D maps for level 3 and above autonomous vehicles. This Karlsruhe-based German company (Atlatec), since its inception in 2014, has focused on bringing a pocket-friendly approach by eliminating the expensive hardware to build HD maps that are tailored for autonomous driving technology.
In an exclusive interview withCEO Dr. Henning Lategahn of Atlatec, he has elucidated some crucial aspects about the need for high-fidelity maps in the autonomous driving industry, the challenges in standardization of map formats, the competitive edge for new entrants over the traditional players, and Atlatec’s future plans.
How are they solving self-driving problem?
Atlatec is consistently advancing to create high resolution, high definition maps for Highly Autonomous Driving (HAD) technology. Dr. Lategahn believes that the high-resolution maps of the environments are essential for vehicles to drive autonomously. HD maps are the easiest way to provide self-driving vehicles with information redundancy, making it one more step safer. Atlatec achieves this with a quite simple hardware set-up of 2 cameras (stereo vision system), 1 GPS receiver, and an IMU (from Xsens). What big giants in the mapping space achieve by using expensive sensors like LiDAR, Atlatec extract the same kind of information and build the same kind of maps with uncomplicated hardware and computer vision.
Why not LiDAR?
LiDAR is quite trending sensing technology in the autonomous vehicle industry and lots of debates are still going on between the tech players on whether to consider LiDAR among the prime set of sensors in the AV sensor suite. Atlatec is firm on its choice of not using LiDAR as it aims at creating highly affordable and precise 3D maps for AVs. Moreover, the company can provide much higher fidelity with just using the camera as a core component, as the camera-based image holds loads of data compared to that from a LiDAR. Henning said that the company does not want to use any component that is expensive. The company strongly relies on computer vision technology and has created a portable sensor box that uses mass manufacturable sensors. Henning also mentioned that for self-driving vehicles it is essential to have a technology that could precisely provide the vehicle localization relative to map structure.
The company is devoted to the post processing software as it is equally difficult to update and maintain the mapping data as it is to record and process it. Atlatec’s software computes the raw data like trajectories, depth maps, and orthophotos from the recorded imagery to a 3D vector format that can be used in the target platforms like self-driving. Also, the software is capable of extracting information like lane geometry, road signs, guideposts, vertical posts, guard rails, and many other objects that are recorded in the mapping run. Dr. Lategahn adds that a high degree of consistency is most important in the areas of self-overlapping and one cannot rely blindly on the GPS. Moreover, to overcome the challenge of time required for the mapping process, Atlatec uses automated data processing and project management software.
Atlatec has taken a step further to make its demonstration maps and data available for the public. The major thought process behind this action is to make the users comprehend the significance of the high-fidelity maps during their real-world tests. The maps available are in a range of formats and can be easily downloaded by any interested parties. The company has the data available for major cities of Europe, United States, Japan, and the areas where level3 and above vehicles are being tested or deployed.
During the conversation, Dr. Lategahn agreed to the fact that mapping companies are publishing maps in their proprietary formats and there is a lack of common standardization. He said that this is one of the major challenges in this sector and will be surmounted with time.
Atlatec is leveraging its computer vision-based technology to accurately locate the position of vehicles even in the GPS-denied environments. The company’s localization maps are best fit with the autonomous driving HD maps. Based on (Simultaneous Localization and Mapping) SLAM technology, it is possible to determine the exact real time position of the vehicles relative to the HD map.
Virtual simulation in the automotive industry has now become more crucial than ever. With the readily available reality-based map database, Atlatec allows companies to test their autonomous system on the complexity of the real world by using its digital twin of the real road. IPG, VIRES, Siemens/TASS, MSC and others are some of the simulation software creating companies with whom Atlatec is directly connected with.
Customers - OEMs and Tier 1s
Touching the major facet of autonomous driving technology, Atlatec is serving the OEMs, Tier-1s and the technology players in the industry with an inexpensive and fast mapping, localization, real world scenario database for simulation. Amongst the giant automakers, Atlatec is serving Ford, AID, ISUZU, and VW; while the tier-1 customers include Bosch, Daimler, Continental, IAV, Visteon, and Denso.
Companies in the mapping industry are competing aggressively to become a worthy asset for the driverless vehicles by offering next generation versions and most precise and high-resolution 3D dynamic maps. Dr. Lategahn believes that “While the big giants like TomTom, HERE Technologies, and Google have the privilege of huge customer base and reliability, there is a huge opportunity in the space for the new entrants to innovate and develop a technology that could serve the AV companies with high level of mapping precision and fidelity”. The one with the best solution in terms of performance and cost will win the race no matter if it is a giant company or a start-up.
In the next 5 years, Atlatec aims at becoming one of the highly recognized mapping and localization companies for the autonomous vehicle industry and envisions to expand its regional mapping coverage to all major cities of Asia, Europe, and the United States.
Conclusively, Atlatec maps the roads with a simple sensor box comprised of stereo camera, GPS, and IMU fitted on the roof of the vehicle, collects the imagery, process the data and extract the objects with the help of manual annotators and artificial intelligence technology to precisely localize the vehicle and generate a computerized perfect map of the area. It runs on a principle of achieving high precision in mapping and localization with cheaper hardware and faster software.