Demonstrating the Connected Intelligent Edge at Embedded World

The international embedded computing community marked its 20th anniversary with this year’s Embedded World conference in Nuremberg. At our booth, innovative companies from around the globe exhibited reference designs and commercial products built around hardware and software from Qualcomm Technologies, Inc. (QTI). In this post, I will describe some of the demos of interest to developers.

The Connected Intelligent Edge
The show was an opportunity to demonstrate the Connected Intelligent Edge, where our 5G and artificial intelligence (AI) technologies help companies connect formerly unconnected things.

Our technology roadmap leads to intelligently connected solutions that help drive digital transformation. The Connected Intelligent Edge is where wireless communications, on-device computing, and low power consumption turn into new business opportunities for embedded developers. It’s where the Internet of Things (IoT) presents us with unique opportunities as it expands across verticals like these:

• Retail — Electronic shelf labels, self-service kiosks, and touchless, biometric check-in stations
• Manufacturing — Ultra-reliable, low-latency control of automated equipment, such as conveyor belts, assisted guided vehicles (AGV), and autonomous mobile robots (AMR)
• Logistics — Precise location tracking, both outdoor and indoor, for real-time tracking of packages and materials in a supply chain
• Energy — Connected energy grids for more efficient production, distribution, and consumption, plus smart meters that allow companies to detect and predict energy surges

Many demos in our Embedded World booth showcased the potential of the Connected Intelligent Edge in key areas including AI at the edge, robotics, and automotive.

AI at the edge – Smart camera demos
Increasing processing power and decreasing power consumption are helping to move AI workloads like inferencing out of the cloud and onto devices at the network edge. Here are some use cases showcased in our booth and partner booths.

Thundercomm – AI Edge Box for operational safety
Suppose you’re responsible for ensuring that employees and visitors wear a helmet or a face covering when they enter your site. Thundercomm helped demonstrate this use case with the use of its EB5 Edge AI Box designed for improving operational safety compliance in industrial and enterprise environments.

The Edge AI Box transforms existing, on-site video cameras into smart cameras running AI at the network edge. The device performs AI analysis on video feeds from as many as eight cameras at a time. It can apply four different AI models (e.g., helmet detection, mask detection, and intrusion alerts) to each feed. After processing at the edge, the EB5 sends results as analytics that appear on a dashboard for plant managers to review.

The EB5 Edge AI Box is powered by the Qualcomm QRB5165 system-on-chip (SoC), which helps support AI, machine learning, heterogeneous computing, enhanced computer vision, and multi-camera concurrency with advanced image signal processors (ISPs).

Lantronix – Vision intelligence
Lantronix has built the Open-Q 5165RB development board around the Qualcomm QRB5165 SoC, and they demonstrated a use case that performs 360-degree facial recognition and detection ideally suited for venues like train stations.

The board performs AI inferencing for people counting, identifying unknown persons, detecting unattended bags, and traffic detection. It can send alerts when passengers stray into safety zones.

eInfoChips - Remote kiosk with AI inferencing at the edge
Can you deliver customer service at the network edge, without continually shuttling data to the cloud? eInfoChips can help developers do this with their Camera Reference Design kit based on the Qualcomm QCS610 processor. The kit combines video capture, face detection, and Amazon Kinesis Video Streams (KVS) for live streaming and alert generation.

The kit uses cameras and face detection to monitor foot traffic. Whenever a visitor steps up to the kiosk, a call to the concierge is triggered. Besides applications in hospitality, the kit demonstrates value in other scenarios:

• Warehouses — Detect when employees enter a restricted area without appropriate personal protective equipment (PPE), and send an alert
• Office buildings — Manage office logistics, deliveries, and security by giving directions
• Retail stores — Help customers purchase products that are not available in the store

The Qualcomm QCS610 application processor offers high performance and power efficiency for on-device AI. It is designed for edge applications like industrial IoT, smart AI enterprise security, home IP cameras, dash cams, body cams, smart displays, and videoconferencing.

Robotics demos
Robotics demos are always a draw at a show like Embedded World, where we showcased smarter and safer autonomous robots for logistics, Industry 4.0, and urban aerial mobility.

Thundercomm – Orionstar service delivery robot
The Orionstar service delivery robot is designed to deliver food and inventory, and to communicate with customers in retail stores. The robot demonstrates an application of AI and edge computing in business scenarios that call for autonomous movement, precise navigation, intelligent obstacle avoidance, and voice communication.

At Embedded World, an Orionstar robot navigated our booth, as if delivering inventory from the back of a store to a gondola for restocking. Traveling a programmed route through the booth without human intervention, the robot avoided people and obstacles.

The Orionstar robot is powered by the Qualcomm SDA845 processor with an AI engine for on-device machine learning and computer vision. Besides retail, its use cases could include hospitality, meal delivery, and customer engagement.

Qualcomm Technologies, Inc. – Qualcomm Robotics RB5 platform
Mounted on a Turtlebot3, the Qualcomm Robotics RB5 platform uses LIDAR for object avoidance, and AI for lane tracking and stop sign detection. It accepts voice commands and wake phrases like “Hey, Turtlebot,” then stays in its lane until it meets a stop sign or an object in its way.

The Qualcomm Robotics RB5 platform uses Robot Operating System (ROS) nodes to direct the Turtlebot3. The application is designed to run lane tracking using AI models trained in ONNX and stop sign detection using OpenCV on frames captured from the camera. It also checks data received from the LIDAR sensor and makes any needed decisions, which the ROS application reads to move the robot and its wheels.

The Qualcomm Robotics RB5 platform is based on the Qualcomm QRB5165 and includes software that manufacturers and developers can use to create consumer, enterprise and industrial robotics products. Ideal use cases include the following:

• autonomous mobile robots (AMR)
• inventory robots
• retail robots
• cobots
• pick-sort-place robots
• cleaning robots
• first-mile robots
• last-mile delivery robots and delivery drones
• prosumer and commercial drones
• industrial and manufacturing robots
• healthcare robots

Thundercomm – Robotic arm
In another appearance of its EB5 Edge AI Box, Thundercomm demonstrated an intelligent robotic arm for industrial applications. In the demo, the robotic arm moves parts to a camera where they undergo quality inspection using on-device AI inferencing.

The box, based on our Qualcomm QRB5165 robotics processor, offers AI, machine learning, heterogeneous computing, enhanced computer vision and ISPs for multi-camera concurrency. The processor is suited to industrial-grade temperature ranges and offers security at every layer.

Automotive Demo
The Snapdragon Digital Chassis is our comprehensive set of cloud-connected automotive platforms for telematics and connectivity, the digital cockpit, and driver assistance and autonomy which empowers automakers to build connected and intelligent vehicles that are safer, customizable, immersive, and continually upgradeable.

Snapdragon Digital Cockpit
The Snapdragon Digital Cockpit is designed for bringing connected, intelligent experiences inside the modern vehicle. Like a central processing hub for connected cars, it supports the driving experience with in-car virtual assistance, contextual safety use cases, advanced audio, graphics and multimedia, and more.

For Embedded World, the demo portrayed the experience of having the Snapdragon Digital Cockpit run eight in-vehicle displays. The fourth-generation automotive development platform drove infotainment services such as navigation, streaming and the digital cluster. Besides running all eight displays, the Snapdragon Digital Cockpit simultaneously ran three operating systems — QNX, Android and Linux — with services spread across all three.

Next steps
The demos we showcased this year at Embedded World are built around commercially available products and kits that you too can build into your company’s innovations. For additional resources

• Qualcomm Robotics RB5 Development Kit
• Qualcomm QCS610 Application Processor
• Snapdragon Digital Chassis
• Thundercomm AI Edge Box
• Lantronix Open-Q 5165RB development board
• eInfoChips Camera Reference Design kit

Snapdragon, Qualcomm QRB5165, Qualcomm Robotics RB5, Qualcomm QCS610, Qualcomm SDA845, Snapdragon Digital Chassis and Snapdragon Digital Cockpit are products of Qualcomm Technologies, Inc. and/or its subsidiaries.