Monday, June 24, 2024

NVIDIA Drive PX Pegasus: Beyond Autonomous Vehicles

Datamation content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.

There are five levels to autonomous driving. Only the fifth level is truly autonomous and doesn’t require a driver at all. Many of the self-driving vehicle products slated to be launched in the next two years are level three and four systems, both of which still require a human driver in the vehicle. And most people expected that the technology to enable a level-five system wouldn’t be ready until early next decade.

Well, apparently, NVIDIA missed that memo and this week announced its Drive PX Pegasus, a level-five system designed for taxis and delivery vehicles. And the Deutsche Post DHL Group announced it was going to be implementing this platform in production for package delivery.

Unlike test systems that could fill a typical four-door sedan, the Drive PX Pegasus is about the same dimensions as a small stack of license plates. This means it could be added to a vehicle without significantly reducing the space available for passengers or packaging.

The applications for a system like this go far beyond cars into general mobile robotics, security and autonomous vehicles of all types including, those that fly.

Currently, when it comes to this level of intelligence, no other company has a production-ready level-five system ready for market. This has big implications for NIVDIA and the markets targeted by this new platform.

Flying Delivery

It is generally easier, from the standpoint of intelligence, to create a flying autonomous vehicle than a ground-based vehicle due to the lack of obstacles and far better sight lines when you are flying. The hard part of flying vehicles is keeping them in the air. This is being addressed on paper by combining large rolling or flying depots where drones could be deployed. All of these requirements could be enabled by a platform like NVIDIA’s. This should substantially reduce the time it will take to bring these solutions to market, placing pressure on regulatory agencies to set and enforce standards and regulations allowing these systems to be deployed.

Amazon has been the most aggressive with bringing a drone-based solution to market, mostly for its own use initially. (But I expect that, to increase its own economies of scale, it will rapidly move to providing delivery services, much like it did with cloud services and AWS.) Given its lead, it seemed unlikely that existing delivery companies could effectively compete. However, a common platform like NVIDIA’s could allow firms like FedEx and UPS to close that technology gap quickly and compete more effectively with Amazon’s coming services.

Other Uses

Security: AI-based deep learning systems are beginning to become viable for large companies, but these systems don’t scale down very well. They certainly wouldn’t be affordable for a smaller business or something like a mall. These systems can look for unusual behavior and track suspicious individuals throughout a complex. This capability is expensive in a security system, but it would be trivial for an AI that can drive a car on the road without a driver to handle these responsibilities. Granted, the configuration would likely be different, but the core technology necessary to drive a car and manage hundreds of real-time elements should also be able to monitor thousands of people.

Manufacturing Floor: Today, real-time management of a manufacturing floor is an involved process. You have quality control elements, individual system maintenance monitoring and massive numbers of diverse parts that need to be tracked and provided real time. The goal is to implement as close to just-in-time as you can to reduce the costs associated with inventories. One of the expected changes over the next decade is the creation of regionalized plants that can build things closer the people who buy them, allowing shorter delivery times, more customization and lower shipment costs. A brain in a box like this could be ideal for this kind of implementation.

Brain in a Box

While initially focused on fifth-level autonomous cars and customers like Uber (which believes that once cars are basically turned into horizontal elevators people will choose to borrow them and not buy them), the potential for the brain in a box that NVIDIA has created with the Pegasus Drive PX system is amazingly broad. I expect this technology to go well beyond cars and eventually into the broad world of security and IoT (in areas like manufacturing) as we start to make everything smarter.

Often vendors significantly overpromise and underdeliver. This may be the opposite case. NVIDIA is focusing its efforts on autonomous cars, but in theory a brain in a box wouldn’t be limited to cars at all. I’m not sure it would truly be limited by anything but NVIDIA’s imagination. This is potentially world-changing.

Photo courtesy of Shutterstock.

Subscribe to Data Insider

Learn the latest news and best practices about data science, big data analytics, artificial intelligence, data security, and more.

Similar articles

Get the Free Newsletter!

Subscribe to Data Insider for top news, trends & analysis

Latest Articles