Autonomous Vehicles, Mapping and Moving On

It is an interesting time to be working in the GIS industry, and I feel grateful for having experienced the past 25+ years immersed in it from an editorial perspective.

DRIVE Map details using radar scan

Sometimes technologies plod along and then a new unencumbered one will arise from left field it seems. I think the pace of development may have accelerated during the pandemic, but the tools were already in place, the visions ready to be put into action before it occurred.

This past month in GIS demonstrated particular developments that will be life-changing for many individuals. R&D is always working on such developments, in a slow and methodical way, but when they reach the collective consciousness, it is then that they become monumental.

One was announced at the NVIDIA GTC conference. NVIDIA traditionally has been known for its blisteringly fast graphics cards. It is now touting an AI-powered world, with new technologies and not to be outdone by a GIS industry that has been toiling for years on such things, it announced a new mapping platform.

NVIDIA CEO Jensen Huang announced the new DRIVE Map platform during his keynote at the annual GTC technology conference on Tuesday.

DRIVE Map is a scalable, multi-modal mapping engine designed to accelerate the deployment of Level-3 and even Level-4 autonomous vehicles, which are being built to operate without human intervention. To achieve higher levels of autonomy, these vehicles require much more detailed maps in order to safely navigate without human assistance. The new mapping platform, DRIVE Map, combines the accuracy of DeepMap survey mapping, a mapping company NVIDIA acquired in 2021 with AI-based crowdsourcing mapping.

This new platform can be used for autonomous driving systems to provide better road information through maps that then provides information to AI to be able to make advanced and safe driving decisions.

This is all in the interest in moving toward safe autonomous vehicles of the future.

According to FutureCar, NVIDIA aims to map hundreds of thousands of miles of roadways in North America, China and Europe to help autonomous vehicle safety navigate with accuracy of 5 centimeters or less. It will provide survey-level ground truth mapping coverage to 500,000 kilometers (310,600 miles) of roadways in North America, Europe and Asia by 2024. The maps will be continuously updated and expanded with data collected from millions of passenger vehicles.

After it acquired DeepMap last year, NVIDIA accelerated development on DRIVE Map. Before it was acquired by NVIDIA, DeepMap’s specialization was fusing crowdsourced images from radar, digital cameras and 3D lidar data collected from passenger vehicles which were the fodder for its high definition maps for autonomous vehicles.

To date, HD maps that are used by self driving vehicles have been a big challenge for developers because they are difficult to keep updated and accessible in real time. The HD Maps do include semantic details not found on standard 2D maps used by millions of drivers everyday for turn-by-turn driving directions. In contrast, highly detailed 3D maps include the exact position of lane markings, road signs, crosswalks, curbs and other infrastructure for added safety.

The building of 3D virtual worlds has long been the province of NVIDIA but now their NVIDIA Omniverse platform shows great promise beyond the entertainment industry.

NVIDIA Omniverse, the platform for artists to collaborate and accelerate 3D work, remains free and is now generally available for GeForce and NVIDIA RTX Studio creators. It’s also where the workflows from DRIVE Map for generating ground-truth training data for deep neural network (DNN) training, testing and validation purposes are being stored.

This was billed as the future of 3D content creation and also how virtual worlds will be built.

Omniverse is very different than a game engine – it is designed to be data center scale. And the biggest breakthrough here is being able to predict and simulate climate change with what NVIDIA is calling Earth Two, the digital twin of the earth.

The portal of Omniverse is USD, Universal Scene Description – essentially a digital wormhole that connects people and computers to Omniverse, and for one Omniverse world to connect to another.

USD is to Omniverse what HTML is to websites. Omniverse is futuristic. The digital twin of the earth will be continuously updated using survey map vehicles along with millions of passenger vehicles.

Omniverse includes automated content generation tools for developers, so that the detailed map can be converted into a drivable simulation test environment that can be used with NVIDIA DRIVE Sim to improve AI-powered autonomous driving software. According to NVIDIA, data such as road elevation, road markings, islands, traffic signals, signs and utility poles are accurately represented in the simulation environment at centimeter-level accuracy.

There is more – NVIDIA has developed a computer simulation environment for autonomous technology developers to have an artificial universe to train robotaxis and self-driving vehicles how to drive in the real world in a simulated environment that’s built using real world data. All testing can take place in the safety of a digital twin environment before venturing out into the real world.

Fleet operators can also rely on the digital twin for a complete virtual view of where their vehicles are, with remote operation as needed.

How long it takes for the actual in-the-trenches work to catch up to the visionaries has always been a question in the minds of those who track this stuff, but in the time I’ve been watching many developments have occurred. We now have complete digital cities, and some partial ones. Greater security, and more accurate data.

It is at this point in history that I take my leave of GISCafe, and move onto real-world applications that don’t require a DRIVE Map to operate, but do require some sensitivity to stay in the driver’s seat. I am grateful that I have watched and reported on the ever-changing landscape of GIS, enjoyed the colleagues I have met over the years, sharing stories and experiences on our various travels. I leave knowing that those who continue will do a wonderful job honing the solutions of the future. As for autonomous vehicles – I just hope they “know” where they’re going!

Tags: laser scanner, reality modeling

Categories: 3D Cities, 3D designs, analytics, Big Data, Building Information Modeling, citizen science, climate change, cloud, data, disaster relief, field GIS, geospatial, GIS, lidar, location based services, location intelligence, mapping, mobile, Open Source, photogrammetry, public safety, remote sensing, resilient cities, satellite imagery

This entry was posted on Friday, March 25th, 2022 at 9:00 am. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.