NVIDIA Jetson TX2 Dana Sheahen ✦ Curriculum Lead - Robotics Software Engineering Nanodegree Program at Udacity ✦ Education: MSCS, BSEE, worked in multiple industries
Jetson TX2 Overview
Jetson: credit card size! ✦ GPU (256 CUDA cores) ✦ CPU ✦ 8G RAM ✦ 32G Memory ✦ I/O (CAN, UART, SPI I2C, GPIO, USB, WiFi, Bluetooth, Camera, Display, GB Ethernet, SATA, …)
AI “at the edge ” Jetson: can do cool stuff! Why? ✦ Real-Time Inference ✦ No waiting for the cloud ✦ Small and low power ✦ Security
AI “at the edge ” Jetson: can do cool stuff! Applications? ✦ Quad-copter ✦ Mobile robots ✦ Industrial robots ✦ Visual systems
Redtail Project on the Jetson https://github.com/NVIDIA-Jetson/redtail Uses DNN for Visual Navigation
Development kit: like a mother board
PCIe & SATA Ethernet SD Card UART HDMI Camera USB2 Mini-USB WiFi GPIO Power
Dev Kit: Ubuntu with GPU
Fun Stuff: Inference example
Getting Started on the Jetson https://developer.nvidia.com/embedded/twodaystoademo DNN Classification, Detection, Segmentation
Fun Stuff - RL Example
Lights Out Robot Workshop @GTC18 on the Jetson https://github.com/udacity/gym-jetson Reinforcement Learning
udacity.com/robotics
Jetson AI Projects to try https://github.com/NVIDIA-Jetson/redtail https://developer.nvidia.com/embedded/twodaystoademo https://github.com/udacity/gym-jetson Thank you! Questions?
Recommend
More recommend