Shubham Shrivastava

Papers

Total Citations

H-Index

About

Shubham Shrivastava is a robotics researcher whose work sits at the critical intersection of uncertainty quantification, trajectory forecasting, and reinforcement learning for autonomous navigation. His most influential contribution, "Propagating State Uncertainty Through Trajectory Forecasting" (2022, 20 citations), addresses a fundamental challenge in modern autonomy: how to systematically carry uncertainty from sensors and perception through the entire prediction pipeline. This work provides a principled framework for treating trajectory forecasting not as a deterministic problem, but as a probabilistic one where uncertainty from upstream components must be faithfully propagated—a crucial insight for safe deployment of autonomous systems. In parallel, Shrivastava's "An A* Curriculum Approach to Reinforcement Learning for RGBD Indoor Robot Navigation" (2021, 6 citations) demonstrates his ability to bridge classical planning with modern learning-based methods. By integrating A* search heuristics into a reinforcement learning curriculum, he enables robots to learn navigation policies more efficiently in photorealistic simulators like Habitat. His research is particularly notable for its practical orientation: rather than treating uncertainty as noise to be ignored, Shrivastava embraces it as a fundamental property of robotic systems that must be modeled and managed for reliable real-world performance.