Papers
28
Total Citations
747
H-Index
12
About
Anirudha Majumdar is a robotics and control systems researcher whose work spans two complementary frontiers: formal methods for robust robot control and the integration of modern AI foundation models into robotic systems. Early in his career, Majumdar made significant contributions to trajectory-centric control theory, developing "funnel"-based frameworks using sums-of-squares programming to provide rigorous stability and safety guarantees for complex robotic systems — work that has garnered hundreds of citations and remains foundational in verified motion planning. His 2017 work on funnel libraries extended these ideas to real-time, uncertainty-aware motion planning in dynamic environments. More recently, Majumdar has pivoted toward some of robotics' most pressing challenges: making foundation models and large language models reliable for physical systems. His highly cited 2024 survey on foundation models in robotics (163 citations) has become an essential reference for the field, while his work on uncertainty quantification — including the KnowNo framework for helping robots recognize when to ask for help — addresses the critical problem of AI hallucination in safety-sensitive applications. His PAC-Bayes Control work further bridges machine learning theory with robot generalization. Across these diverse contributions, Majumdar's unifying theme is trustworthiness: building robotic systems that are not merely capable, but provably reliable.
Research Focus
Key Achievements
Top Papers
- 1Foundation models in robotics: Applications, challenges, and the future163 citations · 2024
- 2Control design along trajectories with sums of squares programming142 citations · 2013
- 3Physically Grounded Vision-Language Models for Robotic Manipulation83 citations · 2024
- 4Robust Online Motion Planning with Regions of Finite Time Invariance75 citations · 2013
- 5
- 6
- 7
- 8Funnel libraries for real-time robust feedback motion planning25 citations · 2017
- 9
- 10