Observe
Fuse telemetry across sensors, controllers, and operating states to understand what the experiment is doing.
Advanced Research & Development // Lab Intelligence
Supervisory intelligence for complex experimental control systems.
Higher-level intelligence for safer experiments, more reliable operation, and better science.
Modern laboratories increasingly rely on software-mediated control systems, yet many instruments still require expert intervention when conditions drift, controllers become unstable, or experimental objectives change.
ViableDx Advanced R&D is developing a supervisory intelligence layer that sits above existing control software. Rather than replacing PID controllers, safety interlocks, or device-control systems, it learns how an instrument behaves, detects abnormal control behavior, helps localize the problem, and guides operator-approved corrective action.
Control architecture
Observe. Diagnose. Guide. Adapt.
Diagnose
Detect instability, drift, and unexpected behavior to identify where the control system is breaking down.
Guide
Recommend operator-reviewed actions based on system state, experimental goals, and operating constraints.
Adapt
Adjust control strategies to fit experiment-specific objectives instead of relying on one fixed operating compromise.
Supervisory intelligence layer
Monitors, reasons, adapts, and guides complex experimental processes.
The supervisory layer operates above existing experimental-control infrastructure. Local controllers continue handling real-time feedback and safety-critical functions, while the intelligence layer reasons across the larger system and longer experimental timescale.
Situation Awareness
Fuse data, detect anomalies, and recognize the current experimental state.
Decision Support
Suggest actions, parameter changes, and better experimental plans.
Adaptive Supervision
Adjust goals, constraints, and control strategies as conditions evolve.
Operational Safeguards
Recognize and respond to undesirable system behavior before it propagates.
Experimental control layer
Real-time control, data acquisition, and device coordination
Feedback Control (e.g., PID, MPC)
Device Coordination
Data Acquisition and Processing
Safety Interlocks
Low-level Autonomy
Current R&D: autonomous laboratory control
Our current research focuses on gain-scheduled PID-controlled scientific instruments, using high-temperature sample environments as an initial testbed. These systems can require manual retuning after hardware changes, and a single poorly tuned operating regime can produce oscillation, offset, or unstable behavior that demands specialist intervention. We are investigating whether a learning supervisory agent can characterize normal operating behavior, detect and localize controller instability, and support faster operator-guided recovery.
Built for heterogeneous laboratory systems
The approach is designed around the software interface rather than a specific furnace or instrument. The longer-term goal is a hardware-agnostic supervisory layer that can support complex experimental systems across neutron science, materials research, quantum systems, and other automated laboratories.
From autonomous devices to intelligent experiments
Instrument automation solves individual control tasks. Supervisory intelligence connects those tasks to experimental context, operator intent, and system-level behavior—helping scientists operate increasingly complex facilities with greater reliability and less manual intervention.
Built for research and industrial partners
Explainable, safety-minded automation for the lab of the future.
We apply the same explainability and evidence-based principles from our healthcare work to the challenges of autonomous science: transparent reasoning, auditable decisions, and human-in-the-loop oversight for high-stakes experimental systems.
Research prototype • Informational only