Process-control engineering, developed through rigorous experiments.
NURU is an independent research lab startup focused on dynamic process modelling, control systems, and the disciplined evaluation of AI-enabled engineering methods.
Purpose
Process industries operate under interacting physical dynamics, hard constraints, disturbances and competing performance objectives. A controller can appear effective under nominal conditions yet fail when conditions shift. Research in this space must combine sound process understanding with systematic testing.
NURU's purpose is to develop and investigate control methods through transparent modelling, simulation, baselines and reproducible evaluation. AI and reinforcement learning are considered alongside established control techniques rather than treated as replacements by default.
Research scope
- Process modelling: dynamic mass and energy balances, simplified equipment models, parameter assumptions and model validation.
- Control engineering: PID, model-predictive control, constrained optimisation and learning-based control policies.
- Reinforcement learning: safe exploration in simulation, reward design, policy robustness and generalisation under disturbances.
- Digital twins: methods to connect equipment specifications, physics-based models and operating data in validated virtual environments.
- Hierarchical automation: control architectures spanning equipment-level decisions and higher-level process objectives.
- Monitoring and diagnosis: fault detection, abnormal-operation indicators and decision-support methods.
How the lab works
Research questions are translated into explicit hypotheses, defined operating envelopes and measurable metrics. Where practical, methods are compared against conventional baselines under the same model, disturbances and constraints. Reproducible code and experimental settings are published when they are ready to support a result.
Safety and technical standards
Simulation is the first evaluation environment, not proof of deployment readiness. Results should include limitations and failure cases. Any work involving a live plant would require appropriate engineering review, validated safeguards, authorised access and a separate deployment risk assessment.
Collaboration
The lab is interested in technically substantive collaboration with engineers and researchers working on process dynamics, automation, optimisation, industrial data and scientific computing. Potential collaborations are not listed as partnerships until both parties have confirmed them.