This blog post was originally published at Qualcomm’s website. It is reprinted here with the permission of Qualcomm.
“It’s amazing how Qualcomm was able to turn the ship on a dime since the last [Embedded World] show. The launch of Qualcomm Dragonwing and the Partner Day event were on point and helpful, showing Qualcomm’s commitment to IoT and edge AI.”
It’s gratifying when one of our partners – in this case, Sigma Connectivity – pays us a compliment like that. For us at Qualcomm Technologies, Inc., Embedded World 2025 in Nuremberg was another chance for us and our partner ecosystem to meet with IoT and embedded developers.
News for developers: Edge Impulse and Dragonwing
The biggest news was Qualcomm’s announcement of intent to acquire edge AI startup Edge Impulse. That underscores our commitment to helping developers implement AI on edge devices powered by Qualcomm Technologies’ processors in areas like security, healthcare, retail and energy, for consumers and enterprises.
The Edge Impulse platform works with Qualcomm Technologies’ edge AI workflow and tools like the Qualcomm AI Hub and Foundries.io. It gives developers even greater access to new models, testing environments and deployment options. More than 170,000 Edge Impulse developers can use the platform for AI capabilities including computer vision, time-series data, audio events and speech recognition. They can build applications at the edge of the network in use cases like manufacturing, anomaly detection, predictive maintenance, and asset tracking and monitoring.
We anticipate enabling developers on Edge Impulse’s platform to build for Dragonwing processors (announced in February) and to take advantage of on-device AI inferencing, computer vision, processing and graphics. Edge Impulse currently supports Dragonwing QCS6490 and Dragonwing QCS5430 processors and is scheduled to support more Dragonwing processors for industrial and embedded IoT applications.
Have a look at the Ask Me Anything (AMA) with Manny Singh of QTI and Zach Shelby of Edge Impulse.
Technology demos
We showcased the demos from 10 partners whose products integrate our technologies to promote edge cloud convergence and on-device AI inference. Here are a few demo videos:
Tria Technologies – Concurrent AI inferencing models
Peter Fenn of Tria Technologies demonstrated the Vision AI-Kit 6490 (Dragonwing QCS6490 Development Kit). With low power consumption and a low thermal profile, the kit can run six different models on two different cameras. Video output from the demo displayed object classification on the left, segmentation on the right and resource utilization along the bottom.
Multi-channel edge AI performance – Qualcomm Innovation Center (QUIC)
Ajit Singh of QUIC showed edge AI performance on two different devices powered by Qualcomm Technologies processors. The Dragonwing IQ-9075 Evaluation Kit (EVK) ran 35 concurrent AI use cases. At its side, the Qualcomm RB3 Gen 2 Development Kit, based on the Dragonwing QCS6490, ran four concurrent AI use cases. Each use case runs its own instance of YOLOv8. These are typical for edge AI use cases, where even legacy cameras can be made smart and you can run AI on every camera feed.
Sign language translation (ASL) – Sigma Connectivity
This demo by Viktor Sannum of Sigma Connectivity used feeds from dual cameras and machine learning to translate American Sign Language (ASL) gestures into closed caption text in real time. The demo ran on a reference platform built around the Dragonwing QCM6490.
Robotic real-time coding – Leda Creative
This demo, conducted by Rex Huang, illustrated a no-code AI platform connected to a robotic arm equipped with computer vision. The arm performed optical inspection of the package and calculated x, y and z- dimensions. It then sent the dimensions to a model running the Qualcomm AIC 100 accelerator board, which automatically updated the code running the robotic arm.
Multi-stage computer vision – Edge Impulse
In this demo, Weiqin Bai demonstrated the application of three different AI models running simultaneously to detect damage, dust/dirt and text legibility. The models ran on an edge AI device powered by the Dragonwing QCS6490.
The feeds from three different cameras offered deep insight into and analysis of three different stages of the manufacturing process.
Additional demos
AI assistant – SECO
An advanced AI-driven technical assistant integrating retrieval-augmented generation (RAG) with an LLM to efficiently deliver precise, contextually relevant responses.
Multi-OS self-checkout kiosk – Sunmi, iterate.ai and AuthSignal
A multi-OS self-checkout solution with biometric payment for enhanced shopping experience, and a Gen AI assistant for troubleshooting and point-of-sale device setup.
Home energy management system – enjoylex, nextDrive
A gateway controller for a home energy management system, demonstrating AI capabilities designed to reduce power consumption, enhance energy efficiency, lower the utility bill and support sustainability.
Open-source, on-device AI development – Thundercomm
Developer-centric demos illustrating highly interactive prototypes and commercialized products, simplifying complex tasks and enhancing project innovation.
AI-enabled IoT edge compute cellular gateway – Lantronix
Edge AI capabilities in an IoT cellular gateway, combining real-time power consumption monitoring with real-world pricing information. Grid operators can direct power and users to cost-optimize consumption.
On-prem AI Appliance Solution – IBM Guardian
LLM technology that prevents malicious queries from being answered when run on Dragonwing AI On-Prem Appliance Solution with the Qualcomm AI 100 Ultra Card.
On-prem AI Appliance Solution – DeepSeek
DeepSeek 70B distilled model with Qualcomm AI Inference Suite, running on Dragonwing AI On-Prem Appliance Solution with Qualcomm AI100 Ultra Card.
Metrics
Our overall activity at Embedded World included:
- The new Dragonwing product brand for the IoT portfolio
- ~100 partner boards, modules and system-on-modules (SOMs)
- 7 Qualcomm Technologies and 10 partner demos addressing multiple industrial use cases
- 50+ partners showcasing Qualcomm Technologies’ products across the expo floor
New hubs and evaluation kits for developers
Besides all the effort we put into spring shows like Embedded World 2025 and Mobile World Congress 2025, we made progress on several other developer-focused initiatives, including the following:
- The Microservices for IoT Solutions hub and Visual Studio Code Extension hub are now publicly available.
- We’ve completed several projects including the Qualcomm QCC74XM and Qualcomm QCC730M Evaluation Kits, FoundriesFactory and Open Source at Qualcomm.
- More updates coming on Dragonwing IQ-9075 Evaluation Kit (EVK).
What’s next?
Join our Discord community to keep up with our ongoing outreach to developers like you who work with Qualcomm Technologies’ products.
On Discord, you can rub elbows with our experts, connect with fellow developers working with our technology and keep your fingers on the pulse of developer-focused news and product updates.
Morris Novello
Staff Manager, Head of Developer Marketing, Qualcomm Technologies, Inc.
