Qualcomm 將覆蓋約130個國家的網絡測試經驗引入機械人平台
Qualcomm Brings Network Testing Expertise From 130 Countries to Robotics
Qualcomm 把從手機晶片開發累積的網絡驗證經驗應用於機械人平台,目標是在極端環境中防止機械人故障。Dragonwing IQ10 可確保機械人上最多40個攝影機、LiDAR 裝置及觸覺感應器的數據及時傳送至主 SoC,平台亦設有獨立安全晶片。Qualcomm 與包括 Neura Robotics 在內的機械人製造商合作,利用數碼孿生模擬器和實地收集的案例,持續重新訓練 AI 模型。

Park Jae-hyun, vice president of Qualcomm Korea, speaks at DETO 2026. (Photo: THE ELEC)
Qualcomm is applying the network validation expertise accumulated through mobile chipset development to its robotics platforms. The goal is to prevent critical robot malfunctions that could occur in extreme physical environments.
Park Jae-hyun, vice president of Qualcomm Korea, outlined the strategy on Nov. 10 during a Seoul National University Industry-Academic Cooperation Foundation session at Deep Dive Into (DETO) 2026, held at the Grand Hyatt Seoul.
"When Qualcomm commercializes a smartphone chip, we test it on local networks in approximately 130 countries worldwide, identifying even extreme corner cases such as communication interruptions and data loss. This helped us become the market leader," Park said. "The same field validation capabilities and processes are now being applied directly to Qualcomm's next-generation robotics platforms."
The company has extended its interoperability testing (IOT) framework to robotics because semiconductor reliability is directly linked to human safety. Humanoid robots have multiple joints and heavy metal bodies. Malfunctions caused by unexpected situations, such as a worker suddenly entering a robot's path or optical sensors misinterpreting reflected light, can result in injuries or substantial financial losses.
Qualcomm has partnered with robot manufacturers, including Neura Robotics, to establish a continuous feedback system that collects corner cases in real-world environments and uses digital twin simulators to retrain AI models.
"We can verify robot operations through simulations in advance, but simulations cannot reproduce every real-world environment," Park said. "Ultimately, robots must be deployed in the field to collect diverse data, which is then used to fine-tune and redeploy foundation AI models to ensure that finished robots can operate properly in commercial services."
With its Dragonwing IQ10 chipset, Qualcomm ensures signal integrity so that data from up to 40 cameras, LiDAR units and tactile sensors installed on a robot reaches the main system-on-chip (SoC) without delay. The platform is designed to ensure that motor joints stop and respond at precisely defined times, even under communication latency or system overload conditions.
The platform also incorporates a dedicated safety chip independent of the main SoC, allowing robots to stop immediately in the event of a malfunction or collision risk. Even if AI processing or the operating system (OS) experiences overload or errors, the real-time operating system (RTOS) and safety domain responsible for motion control and safety remain unaffected. The system uses an isolated architecture to enable immediate emergency shutdown.
Qualcomm's reliability validation capabilities have also been demonstrated in the automotive sector. Its solutions are currently installed in approximately 500 million vehicles worldwide and power more than 90 million digital cockpit displays.
"Over the past 40 years, Qualcomm has introduced low-power, high-performance processor technologies and safety features across the mobile and automotive sectors," Park said. "These solutions have already been validated against the requirements of global manufacturers." He emphasized that this gives Qualcomm a competitive advantage that rivals would find difficult to replicate in a short period.
來源:The Elec · thelec.net