Gupta Lab — Ongoing Research Projects

Clark Atlanta University | Updated: August 28, 2025

1. Physics-Guided Multimodal Data Fusion & Clustering for Enhanced Image Retrieval

Scope: Fuse physics priors with multimodal signals (LiDAR, thermal, RGB, IoT) to boost retrieval precision and clustering. Integrates Rule-set layers, RAG, and LLMs for context-aware search.

Sponsor: AFRL (2023–2026) — $144,400 + $161,000
Outcome: Poster at CVPR 2025, 2 journal + 2 conference publications.

Selected Publications:

K.D. Gupta et al. “Physical Rule Guided CNN for Debris Detection.” SPIE DCS, 2025.
M. Kamal et al. “Physical Fuzzy Rule Based News Article Clustering.” IEEE COMPSAC, 2025.
M.A. Haque et al. “Utilizing Structural Metrics from KGs to Quantify LLM Robustness.” IJDSA, 2024.
K.D. Gupta et al. “Physics Guided Neural Networks with Knowledge Graph.” Digital, 2024.

2. Adaptive Zero-Trust Cybersecurity for Advanced Reactors

Scope: Develop adaptive AI-driven real-time cybersecurity for nuclear reactor control systems, enhancing national energy security and regulatory readiness.

Sponsor: Nuclear Regulatory Commission (2025–2027) — $499,000
Outcome: 1 journal publication.

P. Biswas et al. “Review of Smart Grids with AI & Renewable Integration.” IEEE Access, 2025.

3. PHILIA: Physics Informed & LLM Integrated Autonomy

Scope: Combines physics-informed neural networks with LLMs for explainable, trustworthy, and resilient autonomous systems (UAVs, digital twins, mobility systems).

Sponsor: NASA (2025–2026) — $50,000

4. Trustworthy AI for UAV Controller Security

Scope: Secure UAV controllers using adversarial robustness, anomaly detection, knowledge-graph trust metrics, and AI-driven monitoring.

Sponsor: Georgia Tech (2024–2025) — $35,000
Outcome: 1 journal publication + 1 under review.

M.M. Rahman et al. “UAV Datasets: Segmentation, Classification, Detection, Tracking.” Algorithms, 2024.

5. Assured Autonomy for Human Assisted Drone Operations

Scope: UAV autonomy in GPS-denied or constrained environments, ensuring human-in-the-loop safety.

Sponsor: NASA USRC (2025–2026) — $42,000
Outcome: Student Justin Williams won NASA USRC award.

6. Physics-Guided Digital Twin for Smart Cities

Scope: Build a resilient urban digital twin using multimodal data (LiDAR, thermal, photogrammetry, IoT, weather, social sensing) for real-time monitoring & decision support.

Sponsor: NSF SBIR (under review)
Outcome: Finalist — CIVX Data Challenge; 1 publication pending.

7. MonitorLLM: Continuous Evaluation of LLMs

Scope: Benchmark and monitor LLMs for hallucination, bias, robustness, and reliability using structured metrics (ICR, IPR, CI, SPA, SPI, IMI).

Sponsor: NSF STTR (under review)
Outcome: Poster by Ariful Haque won award at DFRWS USA 2025; Paper Accepted at AAAI Fall Symposium.