Baton Rouge, Louisiana, USA
Bridging artificial intelligence and the physics of the built environment.
01 / About
I build intelligent systems — autonomous robots, deep-learning models, and physics-based simulations — that make the invisible flows of air, energy, and data inside our buildings visible, measurable, and controllable.
02 / Research
Optimizing indoor air quality and HVAC systems with AI and robots. Reinforcement-learning policies (PPO) trade off ventilation, comfort, and energy against measured baselines.
Predicting airflow distribution by combining Explainable Boosting Machines and deep learning with CFD-grounded features — fast surrogates for slow simulations.
A custom ROS 2 mobile robot that logs CO₂, temperature, humidity and airflow with time-sync and mapping — enabling repeatable indoor air-quality surveys across zones.
High-fidelity computational fluid dynamics of particle dispersion, ventilation, and thermal comfort — validated against sensor data from real indoor environments.
03 / Robotics
An autonomous robot I designed and built to map the air you can't see — and trace pollution back to its source.
Fixed air-quality sensors are blind between the dots. So I built a mobile one: a hand-wired robot that drives itself through a room on LiDAR-SLAM, samples CO₂ and PM₂.₅ every two seconds, and fuses those readings into a live map of the invisible. A single robot becomes a whole virtual sensor grid — and then climbs the concentration gradient to find where the pollution is coming from.
Spinning laser scanner that builds the map and localizes the robot in real time (SLAM) — every air reading gets an accurate x, y position.
Tap a marker to inspect the build
It's a closed perception–decision–action loop running on real hardware: sense → localize → fuse → decide where to look next. One self-calibrating robot replaces a rack of fixed sensors, and lays the groundwork for multi-robot cooperative sensing and AI-driven ventilation control.
CO₂, PM₂.₅, temperature and humidity streamed every 2 s — each reading timestamped at the source.
2D LiDAR SLAM maps the room and position-tags every measurement, turning a moving robot into a dense grid of virtual sensors.
Kalman filtering and inverse-distance weighting smooth the sparse trajectory into a continuous, room-wide pollutant field.
Gradient-ascent search climbs the reconstructed field to the emission source — pinned to within ~0.74 m of the true location.
Chahardoli, S., Sarker, A., Lesan, M., Xiao, Y., & Bhattacharya, A.
Indoor Air 2026 (forthcoming)
04 / Publications
Chahardoli, S., Nikoopayan Tak, M. S., Lesan, M., Mousavi, E., & Bhattacharya, A.
Buildings, 16(1), 54
Chahardoli, S., & Lesan, M.
ASHRAE Transactions, 130, 775–783
Chahardoli, S., Khakzand, M., Faizi, M., & Siavashi, M.
Journal of Building Engineering, 53, 104533
Khakzand, M., Chahardoli, S., Niknejad, A., & Khanijazani, T.
Journal of Design and Built Environment
05 / Curriculum Vitae
Hart Howerton
Applying data science, machine learning, and AI to design, planning, and analytics workflows at a leading planning, architecture, and landscape architecture firm.
Dept. of Construction Management, Louisiana State University
Built a ROS-based mobile sensing pipeline; trained Transformer/RNN forecasting and PPO control policies; ran CFD studies validated against sensor data.
Various Universities
Delivered complex technical coursework across 30+ sessions at multiple institutions.
Pahneh Band Consulting Engineers, Tehran
Participated in all phases of design and construction.
Louisiana State University
Optimizing Indoor Air Quality and HVAC Systems using AI and Robots.
Louisiana State University
Thesis: Prediction of Airflow Distribution using EBM and AI.
Iran University of Science and Technology, Tehran
Thesis: Modeling particle dispersion around buildings using CFD.
Malayer University
Project: Research & simulation of houses' energy consumption.
06 / Contact