EMQ Tech Day 2026 | Making Data Flow, by Mr. Lu Xiwen, Carlinx Senior Product Director
Recently at EMQ Tech Day 2026, Carlinx Senior Product Director Lu Xiwen delivered a keynote speech titled "Making Data Flow: How HyperFlow Empowers Intelligent Closed Loops". During the speech, Mr. Lu Xiwen gave a deep-dive on the bottlenecks of data collection in the era of intelligent vehicles, and presented Carlinx's unique HyperFlow solution.
Speech Excerpt:
Breaking Through Industry Pain Points: Achieving Extreme Cost Effectiveness on Vehicle Data Collection
As vehicle electrical architectures evolve, the quantity of signals transmitted has now surged to 7000 or above on a regular basis, causing difficulty in timely collection. Traditional "coarse" data acquisition methods - such as uploading all raw data to the cloud not only waste bandwidth resources but also compromise data's quality due to poor real-time performance and tunnel congestion. To address these challenges, Carlinx implemented three innovations:
Edge Computing & Stream Processing: Through onboard stream processing engine, we offload computing tasks to the edge. Sensitive data is anonymized on the vehicle, and only aggregated results are collected. This approach not only meets stringent overseas compliance frameworks (such as GDPR) but also reduces cloud-side workload.
High Compression Ratio Transmission: High-precision data acquisition at 20ms intervals while employing advanced compression technology, achieving a compression ratio of over 90%.
Real-time Rule Control: Beyond "static" collection methods, HyperFlow supports real-time task dispatch from the cloud, enabling targeted acquisition for specific vehicle models or fault events - preventing invalid data from consuming resources.
Real-world projects shows that Carlinx's solution can reduce traffic consumption by 90% and cloud service costs by 70% or above, providing a highly cost-effective foundation for OEMs expanding overseas.
Empowering AI-Agents: Driving Intelligent Loops on Business Value
“Data's quality determines the level of intelligence”, Mr. Lu Xiwen emphasized. Data acquisition is rather a starting point of enabling AI-driven scenario-based services and intelligent decision-making. By breaking down data silos, Hyper Flow enables data to flow seamlessly across R&D, production, and after-sales phases, giving opportunities to multiple AI agent applications:
Intelligent AI Diagnostics: Traditional diagnostics rely on manual judgment and often suffer from incomplete data coverage. By integrating multi-source logs, CAN signals, and natural language interactions, HyperFlow enables “anomaly-triggered, event-driven acquisition.” This cross-domain data fusion capability allows AI to quickly and precisely pinpoint faults, improving diagnostic efficiency and supporting in-cabin alerts and user feedback loops.
Automated Vehicle Reliability Testing: In the production phase, HyperFlow automatically generate test rules and dispatch them to vehicles. AI then evaluates over 220 functional parameters - including HMI and sensor performance - in real time. This “automation replacing repetitive manual labor” model ensures “zero-kilometer quality” and significantly shortens R&D testing cycles.
Overseas Data Compliance and Flexible Operations: In overseas markets, the challenge lies on the availability of vehicle data. HyperFlow processes sensitive data at the edge, ensuring that critical information never leaves the vehicle. Meanwhile, collection strategies can be adjusted to comply with different regulations, enabling flexible operations within compliance boundaries - ensuring business continuity while mitigating compliance risks.
“Data acquisition is not the end goal - it's about making data truly flow to serve every intelligent business decision,” Mr. Lu concluded.
Looking ahead, Carlinx will deepen its focus on intelligent data services across the entire vehicle lifecycle, helping OEMs transition from “manufacturing excellence” to “intelligent manufacturing excellence”. Together with partners such as EMQ, Carlinx will explore more application scenarios - making every mile more valuable.
