Nvidia announces production ready AD reference platform
Hyperion 8 combines a computer architecture with a complete sensor set for self-driving systems as part of Nvidia’s Drive family. This latest-generation technology is designed to provide the highest level of functional safety and cybersecurity and is supported by sensors from numerous leading suppliers including Continental (long-range radar), Hella (short-range radar), Luminar, Sony (cameras) and Valeo. Drive Hyperion is available today for 2024 vehicle models, Nvidia founder and CEO Jensen Huang announced at GTC.
The production-ready platform is open and modular, so customers can use what they need – from compute hardware and middleware to Level 3 Driving, Level 4 Parking and AI cockpit capabilities. With a functionally secure AI computing platform at its core, Drive Hyperion provides a secure foundation for autonomous vehicle development. During his keynote address at the GTC Keynote, Huang demonstrated a Drive Hyperion 8 vehicle based on a Mercedes model driving autonomously through everyday traffic in Silicon Valley.
Drive Hyperion is scalable: the current standard Drive Orin platform is designed to be upgraded when available with Nvidia Drive Atlan and DriveWorks APIs that are compatible across generations. Two Nvidia Drive Orin SoCs provide the necessary redundancy and resilience, as well as sufficient computing power for Level 4 self-driving and intelligent cockpit functions – even in the case that these functions should only be added later via a software update.
The Drive Hyperion 8 developer kit also includes Nvidia Ampere architecture GPUs. Their processing power provides developers with enough headroom to test and validate new software features, Huang promised. The Nvidia DriveWorks Sensor Abstraction Layer simplifies sensor setup with easy-to-use plug-ins, while the Drive AV software includes the deep neural networks for perception, mapping, planning and control.
The available sensor package for the reference platform includes 12 cameras, nine radars, 12 ultrasonic sensors and a Luminar lidar sensor at the front. By using whichever sensor suppliers are deemed appropriate in conjunction with sensor abstraction tools, autonomous vehicle manufacturers can adapt the platform to their individual self-driving vehicle solutions.
Alongside this, Nvidia unveiled two virtual assistants called Drive Concierge and Drive Chauffeur. Drive Concierge uses Nvidia’s software platform for driver interaction with convenience features – such as reservations, phone calls, online interior monitoring functions and even self-parking. Car designers can use it to implement assistance functions without having to develop their own hardware. The same applies to Drive Chauffeur, which can take over driving tasks on roads. The two virtual assistants are ready for production but can also be modified and supplemented by car manufacturers.
Video: https://www.youtube.com/watch?v=qVyN_chiLeo&t=3s
Related articles:
Omnivision joins Nvidia Drive platform
Virtual reference platform to anchor Linux in the car
Daimler, Nvidia co-develop software-defined vehicle architecture
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
