Shivansh Patel
Papers
1
Total Citations
8
H-Index
1
About
Shivansh Patel is a rising force in robotic manipulation and visuomotor learning, best known for pioneering the real-to-sim-to-real paradigm that bridges the gap between simulation and physical deployment. His flagship work, "A Real-to-Sim-to-Real Approach to Robotic Manipulation with VLM-Generated Iterative Keypoint Rewards" (2025), introduces IKER—a visually grounded, Python-based reward system that leverages vision-language models to iteratively refine task specifications for open-world manipulation. This innovation enables robots to adapt objectives through human-aligned feedback, dramatically improving generalization in unstructured environments. With 8 citations in its first year, the paper has already influenced the robotics community's approach to reward design and sim-to-real transfer. Patel’s contributions lie at the intersection of computer vision, reinforcement learning, and language-guided robotics, offering a scalable framework for teaching robots complex, long-horizon tasks without hand-crafted rewards. His work is particularly notable for its emphasis on iterative, human-in-the-loop refinement—a departure from static reward functions—making it highly relevant for researchers tackling real-world robotic challenges. As a young researcher, Patel is rapidly establishing himself as a key voice in the push toward truly autonomous, adaptable robotic systems.
Research Focus
Key Achievements
Top Papers
- 1