Prime Minister Research Fellow · IISc Bangalore
Rajpal Singh
I am a PhD scholar and Prime Minister Research Fellow at the Department of Mechanical Engineering, Indian Institute of Science (IISc), Bangalore. I work in the DACAS Lab under the supervision of Prof. Jishnu Keshavan.
My research develops data-driven control and learning-based system identification for nonlinear dynamical systems. I focus on Koopman operator theory to construct tractable linear and bilinear representations of complex systems, enabling rigorous model-based control with provable guarantees.
Previously, I received my B.Tech. in Mechanical Engineering (2020) from National Institute of Technology, Srinagar.
Research Interests
01
Koopman Operator Theory
Linear/bilinear lifting of nonlinear dynamics for tractable model-based control.
02
Data-Driven Control
Learning control-oriented models from data with stability and safety guarantees.
03
Robust & Adaptive Control
Prescribed-performance and approximation-free methods for uncertain systems.
04
Autonomous Systems
Control barrier functions, obstacle avoidance, and safe autonomy for UAVs and manipulators.
News
- June 2026 Paper : Generalized Momenta-Based Koopman Formalism for Robust Control of Euler-Lagrangian Systems presented at IEEE ICRA . (arXiv:2601.01971).
- May 2026 Paper : ASACK: Adaptive Safe Active Continual Koopman Learning posted on arXiv (arXiv:2605.09659).
- Apr 2026 Book chapter : Data-Driven Approaches to Real-Time Constrained Tracking Control published in Assistive Robotics, Vol. 2 (Taylor & Francis).
- March 2026 Paper : Generalized Momenta-Based Koopman Formalism accepted to IEEE ICRA 2026.
- Jan 2026 Paper : Deep Robust Koopman Learning from Noisy Data posted on arXiv (arXiv:2601.01971).
- 2025 Paper : Generalized Momenta-Based Koopman Formalism for Robust Control of Euler-Lagrangian Systems posted on arXiv (arXiv:2601.01971).
- 2025 Paper : Adaptive Koopman Embedding for Robust Control published in International Journal of Robotics Research (DOI).
- 2024 Paper : Approximation-free robust tracking control of unknown redundant manipulators with prescribed performance and input constraints published in IEEE Transactions on SMC : Systems (DOI).
- 2024 Paper : Control Barrier Functions for Dynamic UAVs presented at American Control Conference (ACC) 2024.
- 2024 Paper : An Overview of Data-Driven Paradigms for Identification and Control of Robotic Systems published in Journal of Indian Institute of Science (DOI).
- 2024 Paper : Real-time constrained tracking control of redundant manipulators using a Koopman-zeroing neural network framework published in IEEE RAL (DOI).
- 2022 Paper : A Provably Constrained Neural Control Architecture with Prescribed Performance for Fault-Tolerant Redundant Manipulators published in IEEE Access (DOI).
Publications
Journal Articles
Adaptive Koopman Embedding for Robust Control of Complex Nonlinear Dynamical Systems
International Journal of Robotics Research, 2025
Real-Time Constrained Tracking Control of Redundant Manipulators Using a Koopman-Zeroing Neural Network Framework
IEEE Robotics and Automation Letters, 2024
Approximation-Free Robust Tracking Control of Unknown Redundant Manipulators With Prescribed Performance and Input Constraints
IEEE Transactions on Systems, Man, and Cybernetics: Systems, 2024
An Overview of Data-Driven Paradigms for Identification and Control of Robotic Systems
Journal of the Indian Institute of Science, 2024
Preprints
ASACK: Adaptive Safe Active Continual Koopman Learning for Uncertain Systems with Contractive Guarantees
New
arXiv preprint, 2026
Conference Papers
Generalized Momenta-Based Koopman Formalism for Robust Control of Euler-Lagrangian Systems
ICRA 2026
IEEE International Conference on Robotics and Automation (ICRA), 2026
Control Barrier Functions in Dynamic UAVs for Kinematic Obstacle Avoidance: A Collision Cone Approach
American Control Conference (ACC), 2024
Book Chapters
Data-Driven Approaches to Real-Time Constrained Tracking Control of Redundant Manipulators
Assistive Robotics, Vol. 2 — Taylor & Francis, 2026
Projects
-
2025 – Ongoing Safe Continual Active Koopman Learning for Uncertain Systems with Guarantees Ongoing
- Developed online adaptation laws for time-varying Koopman operator models with provable uniform ultimate boundedness guarantees, establishing explicit contraction bounds on parameter error under bounded dynamic drift.
- Integrated the adaptation law with D-optimal active learning MPC that jointly optimizes trajectory tracking and regressor conditioning with control barrier function safety constraints, guaranteeing persistent excitation and robust closed-loop stability.
-
2023 – 2025 Adaptive Koopman Embedding for Robust Control of Complex Nonlinear Systems Completed
- Developed an adaptive Koopman embedding framework combining offline neural embeddings with online adaptation to improve out-of-distribution generalization and robustness to intrinsic and environmental changes.
- Demonstrated superior performance through extensive simulations on coupled pendulums, serial manipulators, and planar quadrotors compared to state-of-the-art alternatives.
-
2024 – 2025 Generalised Momenta-Based Koopman with Linear GESO Completed
- Developed a momentum-based Koopman operator framework for Euler–Lagrange systems using generalized positions and momenta to decouple linear actuation from nonlinear dynamics.
- Integrated a Generalized Extended State Observer (GESO) to enhance robustness against unmodeled dynamics and external disturbances. Validated experimentally on robotic manipulators for robust real-time control.
-
2023 – 2024 Bilinear Koopman Models with ZNNs for Control of Robotic Manipulators Completed
- Developed a deep neural bilinear Koopman-Zeroing Neural Network (ZNN) framework for real-time constrained tracking of redundant manipulators with input and safety constraints.
- Achieved faster computation and improved tracking accuracy compared to traditional Bilinear Koopman-NMPC, with control loop frequency of up to 800 Hz validated in simulation and hardware experiments.
-
2023 – 2024 Control Barrier Functions in UAVs for Kinematic Obstacle Avoidance Completed
- Developed a novel Control Barrier Function (CBF) formulation using collision cones to ensure safe navigation of quadrotors by constraining relative velocities to avoid collision-prone vectors.
- Validated via PyBullet simulations and hardware experiments on Crazyflie 2.1, showcasing real-time QP-based implementation effectiveness in dynamic environments.
-
2022 – 2024 ZNN-Based Control of Underdetermined Systems Completed
- Developed an optimal ZNN control framework for redundant manipulators, guaranteeing prescribed performance and strict input constraint satisfaction.
- Extended the approach to manipulators with unknown kinematics, with performance guarantees for convergence of estimated kinematics.
- Demonstrated robust trajectory tracking, constraint adherence, and fault tolerance through simulations and hardware experiments, outperforming traditional approaches.
Education
-
2020 – Present
PhD, Mechanical Engineering
Indian Institute of Science (IISc), Bangalore
Prime Minister Research Fellow · Advisor: Prof. Jishnu Keshavan -
2016 – 2020
B.Tech., Mechanical Engineering
National Institute of Technology, Srinagar
Teaching
-
Teaching Assistant
Introduction to ROS
MS Ramaiah Institute of Technology, Bengaluru · Mar – May 2025
-
Teaching Assistant
Robotics and AI
MS Ramaiah Institute of Technology, Bengaluru · Jun – Sep 2024
-
Teaching Assistant
Robot Mechanics
NIT Srinagar · Oct – Dec 2022 & Sep – Nov 2023
-
Teaching Assistant
Robotics: Control and Vision
NIT Srinagar · Mar – Jul 2023 & Mar – Jul 2024
-
Teaching Assistant
A Practical Introduction to Data Analysis
IISc Bengaluru · Aug – Dec 2023
Contact
Email
rajpalsingh@iisc.ac.in
Location
Room no 108, Raman Building,
IISc Bangalore, India
IISc Bangalore, India
Google Scholar
View Profile →
GitHub
github.com/Rajpal9 →