What’s Driving the Next Wave of Smart IoT? Insights from Baidu’s Tech Salon

Overview of the Baidu Smart IoT Salon

On April 23, Baidu’s Smart Cloud IoT team hosted the 98th “Smart IoT” technical salon, featuring four senior speakers who presented on IoT platforms, edge computing, visual intelligence, and VR technologies, followed by hands‑on demos.

AIoT Platform and Three Core Missions

Architect Li Le Ding introduced the Baidu TianGong AIoT platform, outlining its three missions: (1) “Connect Everything” – linking devices, data, and compute across space‑time; (2) “Dialogue with Everything” – providing vision, speech, and data services; (3) “Intelligence for Everything” – leveraging AI and big data to create industry‑wide smart applications.

Visual Intelligence and Edge Deployment

Product manager Xu Wei cited research showing 86% of market demand for customized AI, a six‑fold increase in custom models on Baidu AI Open Platform over three years, and that 45% of IoT data is processed at the edge, with over 50% of AI models deployed on edge devices.

He described Baidu’s AI development tools: EasyDL for zero‑code model creation, BML for full‑stack algorithm engineering, and EasyEdge for converting cloud models to heterogeneous edge hardware. The workflow enables remote model deployment through the Intelligent Edge platform.

Edge Computing Demo

Engineer Zhao Xin demonstrated a typical edge‑deployment scenario: instead of manually installing an application on a roadside device, the app is uploaded to Baidu’s cloud console, which then pushes the package to the device, handling updates and start‑up automatically. This illustrates the core edge‑computing loop of device state reporting and remote application delivery.

Virtual Reality Core Technologies

VR lead Chai Xue Zhi explained that VR’s three technical pillars are 3‑D data, multi‑user interaction, and cloud‑rendered graphics. Baidu’s VR stack, built on the Intelligent Cloud and Baidu Brain, offers open capabilities (live streaming, cloud‑VR, multi‑person interaction, 3‑D reconstruction, editing) and a unified platform for content creation across Android, Linux, and other systems.

Senior engineer Yuan Zhenkun showcased the VR Suite, which renders 3‑D scenes, models, panoramas, videos, depth, and roaming, supporting use cases such as VR e‑commerce, virtual house tours, and vehicle showrooms.

Voice‑Semantic Integration with Honghu Chip

Product manager Du Kai described the DuJia voice‑semantic platform, which combines the Honghu far‑field voice chip, speech processing, and semantic parsing. The chip provides real‑time array signal processing, low‑power operation, and offline recognition, while the platform offers Q&A, weather, jokes, translation, and a device‑model‑driven control interface.

Engineer Chen Gaofei led a hands‑on session where participants used the Honghu development board and DuJia service to build an end‑to‑end voice‑controlled demo.

Event Takeaways

The online salon attracted active Q&A from developers, confirming strong interest in practical AIoT, edge, and VR solutions. Baidu announced that future salons will continue exploring emerging technology topics.