AWS re:Invent 2021: 3 IIoT demos from QDN

Thursday 12/9/21 11:39am
|
Posted By Rajan Mistry
  • Up0
  • Down0

Qualcomm products mentioned within this post are offered by
Qualcomm Technologies, Inc. and/or its subsidiaries.

Were you able to attend this year’s AWS re:Invent from November 29 – December 3? Whether you were able to travel to Las Vegas, or attend virtually, Qualcomm Developer Network was there on the expo hall floor to showcase a few industrial IoT (IIoT) demos. We designed them to help get your wheels turning about ways to build capabilities like these into your inventions:

  • On-device AI inferencing
  • Computer vision
  • Edge computing
  • LIDAR
  • 5G
  • Amazon Kinesis Video Streams (KVS)
  • Object/face detection
  • 7-camera concurrency
  • Wi-Fi 6
  • Autonomous drone flight
  • Deep learning
  • The demos were built around SDKs and tools you can download directly from Qualcomm Developer Network, and hardware powered by processors from Qualcomm Technologies, Inc. that support services like AWS Greengrass, AWS IoT device SDK, AWS Kinesis Video Stream, Alexa voice services, and much more! I’ll describe them below, and you can also find a recap of all our AWS re:Invent Demos on QDN.

    Demo 1: Remote kiosk with AI inferencing at the network edge

    Qualcomm® QCS610 Application Processors

    We collaborated with eInfochips, which sells a camera reference design based on the QCS610, on an application that combines a tablet at a kiosk with AI inferencing. It’s an example of putting artificial intelligence in an IIoT device at the network edge to avoid the network round trip for processing.

    This application uses camera and face detection to track traffic in real time. This can be used in a variety of scenarios, such as in a hospitality setting where it can trigger a call to a concierge for virtual support when a guest approaches a kiosk equipped with a device powered by the QCS610 application processor. It uses SSD Mobilenet V2 for face detection and analytics to track the number of calls made. It uses AWS IoT to generate alerts and Amazon Kinesis Video Streams (KVS) to start a video call with the concierge.

    If you’re working on use cases like object/human detection, people counting and license plate recognition, then this demo should get your inventive juices flowing. It’s a good example of taking the compute to the data instead of shuttling the data to and from the compute. The QCS610 application processor offers performance and power efficiency for on-device AI edge applications in industrial IoT, smart AI enterprise security and home IP cameras. See what other developers are building around the QCS610.

    Demo 2: Autonomous robotics navigation and voice activation

    Qualcomm® Robotics RB5 Platform

    Real robots do more than move and lift – they navigate and respond to voice commands.

    This demo is based on the Qualcomm Robotics RB5 Platform, available to you in the Qualcomm Robotics RB5 Development Kit. Here, the kit is mounted on the Turtlebot3 Burger. It uses data from a LIDAR sensor and an interface to an OpenCR board on the Turtlebot3. You can also check out our Autonomous Voice Activated Robot project on Qualcomm Developer Network to replicate on your own.

    We designed this to demonstrate stop sign detection and lane tracking using camera-based AI, plus obstacle detection and avoidance using LIDAR. Also enabled is voice activation to control the robot with spoken commands. The demo showcases functions like AI, deep learning, and computer vision, with the heterogeneous computing power to run it all efficiently. You can connect to it over Wi-Fi 6, or over LTE/5G/CBRS by means of a mezzanine card.

    Are you working on projects like autonomous mobile robots (AMR), inventory robots, retail robots, cobots, pick-sort-place robots, cleaning robots, first-/last-mile robots, prosumer/commercial drones and industrial robots? If so, have a look at additional projects developers have created around the Qualcomm Robotics RB5 Platform using OpenManipulator, Alexa Skills, autonomous exploration/navigation and object detection with TensorFlow Lite.

    Demo 3: 5G and AI in a single unmanned aerial vehicle (UAV) design

    Qualcomm Flight™ RB5 5G Reference Design

    The show provided a chance for autonomous flying enthusiasts to see the Qualcomm Flight RB5 5G Reference Design, available from ModalAI. The reference design is a UAV for applications like disaster relief, asset inspection, mapping, package delivery and film/entertainment. It’s built around the Qualcomm Flight™ RB5 5G Platform and engineered for low power consumption, high performance and low latency.

    The reference design combines two features that you probably haven’t seen in the same package before: high intelligence and long range. You get high intelligence through perception and computer vision with algorithms for obstacle avoidance and self-guidance. Then, you get long range with on-board Wi-Fi 6 (802.11ax) for swarm and drone-to-drone communication. Through an optional mezzanine board, you can add LTE/5G/CBRS and extend your range to the edge of cellular coverage.

    You can use the Qualcomm Flight RB5 5G Reference Design to cut your teeth on heterogeneous computing in a UAV. The Qualcomm Flight RB5 5G Platform includes several dedicated computing engines:

    • Qualcomm® AI Engine, delivering up to 15 trillion operations per second (TOPS) for deep-learning workloads and autonomous flight
    • Qualcomm® Hexagon™ Tensor Accelerator with up to 8 TOPS for accurate edge inferencing
    • Qualcomm Spectra™ ISP for 8K video recording and 4K high dynamic range (HDR) video capture with 7-camera concurrency

    Recap of AWS re:Invent 2021

    For more details on our presence at the show, be sure to visit our dedicated landing page. There you can find information that can help you design for the intelligent edge and invent the Next Big Thing in IIoT.


    Qualcomm QCS610, Qualcomm Robotics RB5 Platform, Qualcomm Robotics RB5 Development Kit, Qualcomm Flight RB5 5G Reference Design, Qualcomm Flight RB5 5G Platform, Qualcomm AI Engine, Qualcomm Hexagon, and Qualcomm Spectra are products of Qualcomm Technologies, Inc. and/or its subsidiaries.