NVIDIA’s GPU Technology Conference (GTC) 2019 is almost here! We don’t want you to miss any IoT content, so in this blog post we’ll share some technical demonstrations that show how emerging technologies, including IoT, edge computing, AI and machine learning, and robotics, will shape our world.
In the AWS Booth (#1221)
Anomaly Detection in The Seattle Spheres
The Spheres is an innovative Amazon building project and home to more than 40,000 plants from the cloud forest regions of more than 30 countries. We’ve collected environmental data since its opening, and a complex MXNet model has been trained to detect anomalies. This demonstration shows time-series anomaly detection using ML inference at the edge powered by the NVIDIA Jetson AGX Xavier module, AWS IoT Greengrass, and Amazon SageMaker Neo. NVIDIA Jetson AGX Xavier runs the AWS IoT Greengrass binary and the cross-compiled MXNet model directly on the device, which is faster and less memory-intensive than running the model natively on the MXNet framework.
Train Machine Learning Models Once, Run Them Anywhere with 2x Performance with Amazon SageMaker Neo
This demonstration shows how AWS customers can use Amazon SageMaker to train a computer vision model in the cloud, use AWS Sagemaker Neo to optimize the ML model for inference on the NVIDIA Jetson Nano AI computer, and use AWS IoT Greengrass to manage a fleet of Jetson Nano devices running ML inference.
In the NVIDIA Jetson Pavilion (#1543 and #1545)
Improving Patient Care with Fall Detection
Designed to improve patient care in elder care facilities and hospitals, this solution monitors human movement and alerts facility staff if a patient falls. This demonstration shows an NVIDIA Jetson TX2 edge device running AWS IoT Greengrass that performs local inference to detect a fall. AWS IoT Greengrass uses the AWS IoT rules engine to send messages to Amazon Simple Notification Service (Amazon SNS), which then sends emails or text messages to facility staff.
Dinosaur World Search
This demonstration envisions a future where vehicles patrol a theme park to identify dinosaur and other rare species and alert scientists of any anomalies or unknown species, all powered by NVIDIA Jetson modules controlled through AWS IoT services. The vehicles travel in predetermined paths, continuously looking for dinosaurs and alert scientists when an unknown species is detected. Scientists can then use Amazon SageMaker Ground Truth to classify new species, retrain the model using AWS SageMaker, and redeploy the improved model using AWS IoT Greengrass.
Developing Robots with AWS RoboMaker
Experience the AWS RoboMaker iterative development, testing, and deployment workflow for robotics applications through a warehouse mapping and navigation application for NVIDIA Jetson devices. Customers can use the AWS Cloud9 development environment for AWS RoboMaker to edit robot application code, test in the AWS RoboMaker simulation using a 3D warehouse simulation environment, and then use AWS IoT Greengrass to deploy the application to a NVIDIA Jetson device.
Stop by the AWS booth or the NVIDIA Jetson pavilion to learn more about our collaboration to bring AI, machine learning, and robotic development to millions of edge devices.