Publications & Research

Academic publications, theses, and research project contributions focusing on multimodal deep learning, trustworthy AI, and sensor data interpretation.

Towards Trustworthy and Explainable AI for Perception Models: From Concept to Prototype Vehicle Deployment

Conference Paper | IEEE International Conference on Intelligent Transportation Systems (ITSC 2026)

Investigated the structural resilience, interpretability, and adversarial robustness of multimodal attention mechanisms integrating LiDAR point clouds and camera data. Developed a complete pipeline deployed on a prototype vehicle to evaluate out-of-distribution edge cases and sensor degradation.

T. Beemelmanns, S. Sharifi, M. Mehrotra, A. Choudhuri, L. Eckstein (2026). Towards Trustworthy and Explainable AI for Perception Models: From Concept to Prototype Vehicle Deployment. In IEEE International Conference on Intelligent Transportation Systems (ITSC), Naples, Italy. [arXiv:2605.16087].
Read Paper on arXiv

Robustness of Multi-modal Transformer 3D Object Detectors

Masters Theis | Institute for Automotive Engineering (ika), RWTH Aachen University (2024-2025)

Designing of a custom-made steering assembly for an All-Terrain Vehicle

Bachelor's Thesis | Birla Institute of Technology, Mesra (2019)

Conducted finite element analysis (FEA) and structural optimization for mechanical sub-assemblies to improve load-bearing efficiency and operational life under dynamic stress conditions.

Mehrotra, M., Patel, S., Sharma, A., and Mishra, S. K., Designing of a custom-made steering assembly for an All-Terrain Vehicle, National Conference on Recent Ad- vancement in Mechanical Engineering under TEQIP III at Ranchi on February 18-19, 2019, pp. 102–112, e-ISBN: 978-93-8830-572-3.