Due to our team’s recent breakthroughs in physical AI and AIoT, this website and the websites of all 64 portfolio companies are being substantially upgraded to incorporate our Physical AI and AIoT engines and sensing technologies. This overhaul will be completed by Oct. 31, 2026.
This hub page and all of its leaf pages have been updated.

Physical AI & AIoT Research Initiatives

Physical AI & AIoT Research

The joint research initiative across Aperture Venture Studio, GAO RFID Inc., and GAO Tek Inc. advances six core domains of Physical AI and the Artificial Intelligence of Things (AIoT). By integrating robust identification protocols, predictive intelligence, closed-loop physical control, operational digital twins, distributed multi-agent systems, and continuous runtime verification, our research bridges digital decision-making and real-world execution.

Below is a list of our research programs.

Industrial Identification, Sensing and State Estimation

Establishing a foundational understanding of physical assets, equipment, personnel, and dynamic operating conditions. This program encompasses multimodal sensor fusion (RFID, UWB, BLE, and computer vision), AIoT-driven inventory tracking, and real-time data reliability intelligence to ensure decision systems act on accurate physical telemetry.

Multimodal Industrial Identification, Localization and State Estimation: Fusing RFID, UWB, BLE, and visual inputs to resolve identity, spatial position, and state across complex industrial environments.
AIoT-Based Inventory, Asset Movement and Process Understanding: Inferring operational events, transfers, and inventory state transitions despite missing or noisy sensor reads.
Industrial Sensor Reliability and Data-Quality Intelligence: Detecting sensor drift, calibration errors, and communication gaps before unreliable data skews downstream AI decisions.

Distributed Intelligence, Robotics and Physical Logistics

Scaling isolated AIoT devices into resilient, multi-agent networks capable of physical coordination, material handling, and adaptive manufacturing.

Resilient Edge AI and Distributed Industrial Intelligence: Allocating compute and inference dynamically between sensors, gateways, and edge servers to maintain resilience during network outages.
Multi-Agent Coordination for Industrial Operations: Managing task allocation, asset contention, and workflow negotiation across autonomous agents.
AI-Assisted Material Handling, Warehouse Robotics and Physical Supply-Chain Execution: Directing autonomous mobile platforms and material handling equipment using real-time RFID/UWB spatial context.
Adaptive Robotic Assembly and Machine-Tool Assistance: Enabling machine tools and robotic manipulators to adjust dynamically to part variations and process disturbances.

Equipment Health, Inspection and Quality Intelligence

Delivering commercial intelligence through predictive diagnostics, multimodal visual/acoustic inspection, and process root-cause analysis prior to full autonomous execution.

Transferable Predictive Maintenance and Equipment Diagnostics: Learning equipment degradation patterns across operating regimes with minimal failure datasets.
Multimodal Industrial Inspection and Defect Detection: Combining visual, thermal, acoustic, and vibration telemetry for comprehensive component and infrastructure defect recognition.
Quality Prediction and Root-Cause Analysis in Industrial Processes: Linking upstream physical measurements with downstream quality metrics to identify root causes and enable proactive interventions.

Verified AI Decisions and Physical Control

Translating autonomous recommendations into controlled, measurable, and safe physical outcomes.

Verified AI Agents for Industrial Device and Equipment Control: Translating high-level operational intent into constrained, safe execution sequences via gateways and industrial APIs.
Constraint-Aware Adaptive Control and Industrial Optimization: Balancing operational throughput, energy efficiency, and equipment wear within strict physical setpoints.
AIoT and Physical AI for Building and Energy-System Operation: Optimizing complex utility, HVAC, compressed air, and thermal storage infrastructures under dynamic conditions.

Verification, Human Oversight and Emerging Connectivity

Ensuring runtime safety, explicit constraint enforcement, and communication-aware control across all physical deployments.

Runtime Assurance, Human Oversight and Physical Outcome Verification: Implementing continuous safety monitoring, fallback states, and independent outcome confirmation.
Bounded Formal Verification and Foundation-Model Action Assurance: Checking foundation model action outputs against strictly specified rules and permission boundaries.
Wireless Sensing and Communication-Aware Physical AI: Adapting physical control policies to dynamic wireless channel conditions, latency constraints, and link reliability.

Industrial Digital Twins and Simulation

Constructing real-time synchronized representations of machinery and leveraging synthetic data to accelerate physical deployment.

Operational Digital Twins for State Estimation and Action Evaluation: Synchronizing physical models with live sensor feeds to estimate hidden states and pre-evaluate physical actions.
Simulation-to-Reality Transfer and Synthetic Industrial Data: Utilizing synthetic environments to train models for edge cases, reducing physical experimental overhead.