Imagine using a smartphone to summon a driverless car with one click: A fully automated car based on the DRIVE Pegasus AI supercomputer will open a new 'mobile as a service' era.
Just click on the button on your smartphone and you will be off. Daimler, Bosch and NVIDIA have teamed up to work on fully automated driverless vehicles on city streets. The collaboration will be more than just driving. change.
Although the world's billion-dollar cars travel 10 miles per year, these vehicles are idle most of the time and occupy valuable floor space when parked. They are often trapped on congested roads while driving. Above. Mobility services will solve these problems that plague urban areas, unlock potential and completely change the way we travel.
On a global scale, we see companies such as Uber, Lyft, Didi and Ola are rapidly adopting new mobile services. But the limited number of available drivers may limit their continued growth.
And driverless cars can be the solution – they are equipped with a large number of sensors, powered by extremely energy-efficient supercomputers, and run AI software that can be used as a virtual driver.
On July 11, Daimler, Bosch and NVIDIA announced a collaboration to promote the development of the L4 and L5 autonomous driving in the automotive industry – a car that can drive autonomously.
Mobility services based on autonomous vehicles have many advantages. These AI-integrated vehicles will improve traffic flow, enhance safety, and provide better mobile services. In addition, analysts predict that unmanned driving on the go The cost of the car will be only 17 cents per mile. Office workers will be able to work on the road to recover a loss of productivity worth about $99 billion a year.
While improving transportation convenience and reducing costs, this capability inherently presents tremendous opportunities. KPMG said that by 2030, driverless cars and services will develop into a $1 trillion industry.
To seize the opportunity, well-known car brands need to incorporate the latest technology into their current work. Our AV computing platform NVIDIA DRIVE will help them combine these technological breakthroughs with their own architecture – including deep learning, sensor fusion, image recognition, cloud computing Wait.
Our partnership with Daimler and Bosch will give full play to their respective advantages. NVIDIA is a leader in AI and autonomous driving platforms. Bosch is the world's largest Tier 1 automotive supplier with expertise in hardware and systems. Mercedes Daimler, the parent company of Benz-Benz, brings expertise in the field of complete vehicles, and its global brand is synonymous with safety and quality.
Realizing road intelligence
We are coping with huge challenges. Faced with pedestrians, cyclists, traffic lights and other vehicles, even good human drivers are under pressure when driving on crowded city streets.
In this chaotic, unstructured environment, the demand for computing power has increased dramatically. Bosch says that only one camera can generate 100 gigabytes of data per kilometer.
Now imagine a fully automated vehicle or a self-driving taxi with a high-resolution camera, lidar and radar sensors around the body that can sense objects from a distance. At the same time, they will also be used with a variety of The combination of color observation, measurement distance, and motion sensor detection under various conditions. These systems create levels of diversity to improve safety and provide redundancy to provide backup in the event of a failure. However, this large amount of information Need to be deciphered, processed and put into use by multi-layer neural networks in almost instantaneous.
NVIDIA DRIVE delivers high performance to run a variety of deep neural networks simultaneously, ensuring that vehicles are safely transported in urban environments.
To run dozens of complex algorithms at the same time, it requires extremely high computational performance, so that it can be executed within a few milliseconds, so that the car can drive safely and comfortably.
Daimler and Bosch choose DRIVE Pegasus
NVIDIA DRIVE Pegasus is an AI supercomputer designed for self-driving cars that offers 320 TOPS (teraflops per second) performance to handle these diverse and redundant algorithms. It only has license plate size and performance. At the same time running six desktop workstations.
This is the most energy-efficient supercomputer ever – running one trillion operations and consuming only one watt. By minimizing energy consumption, we are able to expand our operating range.
Pegasus' architecture is designed to ensure security and high performance. This car-grade, feature-safe production solution uses two NVIDIA Xavier system-on-a-chip and two next-generation GPUs designed for AI and visual processing. This jointly designed hardware and software platform has for the first time in history meets the requirements of the ASIL-D ISO 26262 standard, which is the industry's highest level of automotive functional safety. Even when a fault is detected, the system can still operate.
From the car to the cloud
NVIDIA AV solutions go beyond the capabilities of traditional cars. The NVIDIA DGX AI supercomputer for data centers trains deep neural networks to deliver superhuman perception. The new DGX-2 features 2 petaflops for deep learning The time, space, energy and cost required are much lower than the CPU server.
After training on a powerful GPU-based server, the NVIDIA DRIVE Constellation AV simulator can be used to test and validate the complete software 'stack' that is ultimately placed inside the vehicle. This high-performance software stack covers autonomous vehicles. Everything from object detection, deep learning and computer vision, to map localization and path planning, everything runs on DRIVE Pegasus.
In the next few years, DRIVE Pegasus will play a key role in helping automakers meet their surge demand. Market research firm IHS Markit says that by 2040, the 'mobile as a service' industry will purchase more than 10 million vehicles, much higher than 300,000 vehicles in 2017.
'The collaboration with Bosch and Daimler shows that the NVIDIA DRIVE Pegasus architecture meets the key needs of automakers in responding to the autonomous driving challenge,' said Luca De Ambroggi, senior research director of IHS Markit AI, 'NVIDIA will be AI hardware, software The integration platform and the combination of simulation and verification tools create more value for AV development.'
a thriving 'mobile as a service' ecosystem
The NVIDIA DRIVE ecosystem continues to expand in all areas of autonomous driving, from autonomous taxis, trucks to delivery vehicles. More than 370 companies have adopted the DRIVE platform. Today, our partnership with Daimler and Bosch will create Innovative unmanned vehicles and services will not only optimize the city streets, but will also bring changes to our lives.