We develop control-theoretic methods for enforcing safety in nonlinear constrained systems. A central theme in our work is the design of safety-embedded control architectures, including barrier state (BaS) methods that integrate safety directly into system dynamics and feedback design. Our research studies multi-objective safe control (stabilization, tracking, regulation, etc.), safe optimal control and nonlinear safe feedback synthesis with theoretical guarantees and scalable algorithms. We are interested in both theory and computation: from new formulations for safe multi-objective control to practical algorithms for robotic autonomy in cluttered or uncertain environments.

We study optimization-based control and planning methods for autonomous systems operating under model uncertainty, disturbances, and safety constraints. Our work includes robust and safe trajectory optimization, MPC, sampling-based control, and differentiable optimization. A key objective is to design methods that remain computationally practical while improving robustness, safety, and performance in complex robotic environments.

We develop learning-based, adaptive and data-driven control methods for systems whose dynamics, environments, or operating conditions change over time. Our interests include safe adaptive control, learning-enabled control under uncertainty, Gaussian-process-based safe control, and online adaptation in high-performance settings. Rather than treating learning as a black box, we aim to integrate adaptation and data-driven modeling with control structure, safety, and real-time decision making.