【Technical Deep Dive】Integrating Edge AI and rPPG Technology—PSIE Multimodal Inference Engine Prototype Expected for Release in Q2 2026

Daviundler Research Lab today unveiled the technical specifications of its "Psychological State Inference Engine" (PSIE), demonstrating its technological leadership in the field of Edge AI.

The system utilizes an advanced "Multimodal Sensor Fusion Architecture." Its primary competitive advantage lies in the simultaneous integration of visual features, vocal emotions, environmental parameters, and the highly innovative rPPG (Remote Photoplethysmography) non-contact sensing technology. Through imaging alone, the system can precisely infer an individual’s pulse and heart rate variability (HRV), allowing the AI to obtain core physiological data without any interference or physical contact with the user.

In terms of technical architecture, we have implemented an original "Parent-Child Model System." The "Parent Model" handles generalized, large-scale multimodal baseline labeling, while the "Child Model" utilizes Few-Shot Learning technology to perform rapid fine-tuning and continuous learning for specific individual users. Combined with our newly developed RLAF (Reinforcement Learning From Auto Feedback) mechanism, the PSIE engine becomes increasingly accurate in mastering the psychological habits and reactive logic of a specific individual over time, truly achieving the ideal of "Personalized AI."

To ensure data privacy and low-latency response, we have collaborated deeply with Qualcomm to deploy PSIE on the Snapdragon Edge Computing Platform. This means the majority of psychological state inferences are completed locally on the device (On-Device AI). Sensitive biometric data never leaves the terminal, meeting the stringent privacy protection requirements of high-end markets.

Currently, the development team has entered the final stages of system integration and testing. We officially expect to produce the first PoC (Proof of Concept) prototype by the second quarter of 2026. This prototype will demonstrate real-time psychological state mapping, dynamic environmental adaptation strategies, and the data transfer process of our "Soul Recall" technology, opening up diverse commercial possibilities for the future—ranging from intelligent care and educational technology to high-end autonomous robotics.