Enabling Smart Robots with Low-Power Computer Vision
What does it take to create smarter robots and autonomous machines? How can parallel processing and embedded systems enable a new range of products and business models? With recent advances in neural networks at NYU, Berkeley and elsewhere, companies are building GPU-accelerated vision and voice recognition systems, and are also planning systems around neural networks.
The first part of this talk will present NVIDIA’s work with neural networks and more traditional computer vision systems, enabling massive speedups on parallel hardware. The second part of this talk will explore Jetson, a developer board created to serve a perceived need for low-power autonomy and vision systems. Interesting depth-imaging and camera peripherals will be discussed, as well as the rewarding challenge of a using a coherent software and documentation ecosystem to address the needs of academics, homebrew enthusiasts and product developers.
This talk will be preceded by the brief CNSV Annual Meeting.
About the speaker, Barrett Williams of NVIDIA
Barrett Williams is Technical Marketing Manager and Robot Hacker at NVIDIA in Santa Clara. He began building robotic prototypes with a vintage PowerMac and a Lego robotics kit. After writing a tech column at Yale, he switched tracks from economics to electrical engineering at Columbia. Today you’ll find him building robots with 3D-printed parts, or helping others to do so.
Prior to NVIDIA, Barrett ran web infrastructure and development for a tutor-matching site, and designed components for Meta’s AR glasses. When he’s not obsessing over robots, he’s probably biking or skiing somewhere in the Bay Area or Lake Tahoe.
Location: Hewlett-Packard Co. - 3000 Hanover Auditorium
3000 Hanover St., Palo Alto, CA 94304
View Map & Directions