NVIDIA GTC 2021 kicked off Monday. Computer-generated art and aphorisms danced across the screen. AI and AVI-composed music performed by the London Symphony Orchestra accompanied. Climaxing with the phrase ‘I am AI – brought to life by NVIDIA, Deep Learning and Brilliant Minds, Everywhere’.
NVIDIA CEO, Jensen Huang, clad in his iconic leather jacket, appeared. He was flanked by a stainless-steel stove, bas-relief sculpture and a granite countertop. For an hour and a half he announced many new technology products.
Huang was transported into NVIDIA-generated Omniverses. He lauded NVIDIA’s automotive, autonomous vehicle, automotive production and future robotic technologies.
GTC 2021 offered a cacophony of intelligent sessions. Topics included aspects of AI, autonomous driving theories and the future of mobility.
“AI and 5G are the ingredients needed to kick start the fourth generation of the industrial revolution – where automation and robotics can be deployed to the far edges of the world. The one more miracle we need is the metaverse which is a virtual twin of the world we live in,” says Huang.
NVIDIA is expanding the reach of its platform to reach new markets. Omniverse is a platform to create and simulate virtual worlds, also known as digital twins.
BEEM Me Up to The Virtual BMW Factory
Let’s take a look at the BMW factory, says Huang, and then Milan Nedeljkovic, Member of the Board of Management of BMW AG production appears on the screen and says: “BMW is setting the standard for innovation and flexibility using the NVIDIA Omniverse and AI a new era of digitalization of automobile production.”
“We are inside the digital twin of a BMW assembly line for the first time. We are able to have our entire factory in simulation. Global teams can collaborate on different software packages. The ability to work in a perfect simulation revolutionizes BMW’s planning processes,” says Nedeljkovic.
BMW regularly reconfigures its factories to accommodate new vehicle launches. At one point, Huang is transported in the digital world of the factories.
In a demonstration, a planning expert “wormholes” into a simulation in a motion capture suit moves as a factory worker, while another expert adjusts the line design in real-time for worker economics and safety.
BMW is also employing a fleet of intelligent robots to move goods around the factory, controlled, designed and monitored by NVIDIA’s Isaac robotic platform with updatable software.
“The complexity of our production network requires BMW to constantly innovate. I’m happy about the tight collaboration between our two companies. NVIDIA Omniverse and NVIDIA AI give us the chance to see all 31 factories of our production Network at BMW,” says Nedeljkovic. “These new innovations will reduce planning times, improve flexibility and produce 30% more efficient planning processes.”
“We’re building an AV platform from end to end from AI chips, computers, sensor architecture, data processing, mapping to developing the driving software. We are also creating the simulator and digital twin,” says Huang. He notes that NVIDIA built the end-to-end AV service with Mercedes-Benz.
TOP Automotive & Autonomous NVIDIA Tech Announcements
The car industry has become a technology industry. The car is not just a product to sell but an installed base of tens or hundreds of millions to build upon— creating billions of services opportunities, says Huang. He points out that some people are calling TOPs the new horsepower.
NVIDIA officially announced automotive product upgrades, chips platforms and services.
· ORIN will be a one central processing computer, the cluster, infotainment, passenger interaction and monitoring. Plus the 3D surround shows the riders what the car is seeing and doing.
· 8th Generation DRIVE Hyperion AV Platform streamlines data collection, AV development and testing. The platform now includes DRIVE Software source access to help accelerate the path to production for DRIVE customers.
· NVIDIA DRIVE Atlan offers greater computer power than in most robotaxis today with 1,000 TOPs on one chip. Atlan is also the first SoC to be equipped with an NVIDIA BlueField data processing unit (DPU) that will be available in a couple of years The software is compatible with previous DRIVE compute platforms.
· NVIDIA DRIVE Sim is now powered by NVIDIA Omniverse. DRIVE Sim enables high-fidelity simulation using NVIDIA RTX, Omniverse and AI. It can generate datasets or provide a virtual proving ground.
Back to the (Autonomous) Future
Autonomous driving is not easy. There are theories behind how the self-driving software runs, how it is simulated and what will happen in the future.
In a presentation, ‘Game-Theoretic Viewpoint in Self-driving and in Machine Learning’, Drew Bagnell, Chief Scientist, Co-Founder, Aurora says: “We have to understand what all the actors are doing. You need to react like a safe human.”
He showed various models for making autonomous driving decisions in one use case, software anticipates that a bicycle will come into the lane to go around a car parked on the side of the road. Bagnell also negates the media’s contention that every human take-over of a self-driving vehicle on the road is bad.
Bagnell predicts: “I see more and more AV multiple actors interacting forming an equilibrium solution.”
We will soon live in a world driven by data, connectivity is the drivetrain of the future.
Raquel Urtason, formerly Chief Scientist-at Uber and professor at the University of Toronto presented, ‘Next-Gen Simulation Testing and Training self-driving Cars at Scale’, giving some technical tips.
“We’re still very far from having a commercial solution for self-driving. We need self-driving 2.0. We need to go from the robotic centric approach to an AI approach. We need more sophisticated algorithms that can solve long-tail scenarios. We need new ways to test and verify our systems,” says Urtason.
She went through several components of what she called the ‘holy grail’ of autonomous simulation. “There needs to be real-time high fidelity closed-loop simulation for testing and training autonomous systems and failures that identifies failures and adversary simulations.”
Hildegard Wortman, Member of the Board of Management for Marketing and Sales Audi AG, presented Audi’s vision for the future.
“Premium mobility is not just getting from A to B. Driving is an attitude. It is a statement and being part of a more sustainable world,” says Wortman.
In the future, the car will no longer be designed from the outside-in starting from the car body e but from the inside out. The development and design process starts with who the new model made is for and why. The approach builds upon specific use cases for each model range and the degree of automated driving, says Wortman.
“We will soon live in a world driven by data, connectivity is the drivetrain of the future. Everything will be connected” notes Wortman, “The car of the future is fully electric and fully autonomous.”
Huang gave a glance into the future technology coming from NVIDIA, “Eventually everything that moves will be autonomous.”