Nvidia looks toward AI-powered future with Jetson TX1 debut

11 Nov 2015 | Author: | No comments yet »

Kespry and NVIDIA Demonstrate Deep Learning for Commercial Autonomous Drones.

Kespry, a commercial drone system company, today demonstrated a prototype drone that uses NVIDIA artificial intelligence technology to recognize objects.NVIDIA may be a graphics powerhouse when it comes to add-in boards for desktops and integrated graphics for notebooks, but the company isn’t putting all of its eggs into one basket. The state-of-the-art Kespry prototype uses an NVIDIA Jetson TX1 module for deep learning, which offers complex algorithms to make autonomous devices more intelligent. It also tiptoes into supercomputing efforts, automotive technology (the company’s Tegra X1 powers the 2016 Audi TT’s “virtual cockpit” gauge cluster), and even SoCs for mobile devices.

The prototype is based on the Kespry Drone System that is in use by customers in the materials, mining and construction industries. “This technology has great potential for the commercial drone market,” said Paul Doersch, Founder and CEO, Kespry. “Today, Kespry customers already use aerial data gathered by our drones to calculate distances, sizes, and volumes. NVIDIA is expanding its reach into new markets with Jetson TX1, a new platform that has a footprint smaller than a credit card and is destined for autonomous devices. Aerial drone leader DJI was already moving down this past using NVIDIA hardware — a Tegra K1 SoC to be exact — with Manifold PC which pairs with its Matrice 100 drone. It’s not about frames per second; instead, Nvidia is focusing on high performance computing tasks – specifically computer vision and classification – in a platform that uses under 10 Watts. While these Raspberry Pis, BeagleBones, and router-based dev boards are great for running Linux, they’re not exactly very powerful. x86 boards also exist, but again, these are lowly Atoms and other Intel embedded processors.

But NVIDIA isn’t exclusively targeting big companies or startups with lots of cash to throw around: if you want to buy a single Jetson TK1, you’ll be able to do that… pricing for individual units just hasn’t been determined yet. Well, the 3.4″ x 2″ module features A Tegra X1 ARM Cortex-A57 processor with 256-core NVIDIA Maxwell graphics, 4GB of LPDDR4 memory, 16GB of eMMC storage, 802.11ac WiFi and Bluetooth, Gigabit Ethernet support, and a 400 pin board-to-board connector for hooking it up to other hardware. But where NVIDIA see its real advantage, at least compared to larger players like Intel is a performance-per-watt (the platform consumes less than 10W), which it has outlined in the following charts: Jetson TX1 is designed to use deep neural net training and applies what it “sees” in the field with its cameras to identify objects and allow for a new superclass of autonomous products ranging from drones to robots that can navigate disaster areas to dangerous for humans. Image classification is one of the most computationally intense tasks out there, but for autonomous robots, there’s no other way to tell the difference between a cyclist and a mailbox.

That’s because the Tegra X1’s high-performance graphics offers the kind of parallelized processing power necessary for the job, while consuming less than 10 watts of power (so you don’t need room for a big fan, heat sink, and battery on your drone). As we’ve seen, autonomous vehicle technology is hot these days, with Tesla Motors recently fielding its Autopilot beta for Model S sedans to predictable results. To do this on an embedded platform, you either need to bring a powerful general purpose CPU that sucks down 60 or so Watts, or build a smaller, more efficient GPU-based solution that sips a meager 10 Watts. Even the world’s largest automobile manufacturer, Toyota, has finally seen the light by investing $1 billion into the technology for future vehicles. “Jetson TX1 will enable a new generation of incredibly capable autonomous devices,” said Deepu Talla, VP and GM of NVIDIA’s Tegra unit. “They will navigate on their own, recognize objects and faces, and become increasingly intelligent through machine learning. But you can use a higher-performance computer to develop a routine, submit it to the Jetson TK1 and it can put those algorithms into practice in real-world environments to avoid obstacles or respond to people, animals, or other objects even if it’s never seen them before.

It will enable developers to create industry-changing products.” A Jetson TX1 developer kit will be available on the retail market (Amazon, Newegg, Microcenter, etc.) for $599, while educational customers will be able take advantage of a much lower price tag of $299. This includes Nvidia Visionworks, OpenCV, OpenVX, OpenGL, machine learning tools, CUDA programming, and everything you would expect from a standard desktop Linux box. This deep learning task uses a neural net trained to differentiate between objects; pedestrians, cars, motorcycles and cyclists if the TX1 will be controlling an autonomous car, or packages and driveways if the TX1 will be used in the first generation of delivery drones.

Here you can write a commentary on the recording "Nvidia looks toward AI-powered future with Jetson TX1 debut".

* Required fields
All the reviews are moderated.
Twitter-news
Our partners
Follow us
Contact us
Our contacts

dima911@gmail.com

ICQ: 423360519

About this site