Practical Autonomy

Optimization and learning tools for autonomous systems operating in uncertain, real-world environments.

Robotics

My work on robotics studies how safety-critical control methods can be integrated into autonomous decision-making and motion planning. This includes online construction of control barrier functions for manipulator safety, safety-aware optimal control for inspection robots operating in constrained and hazardous environments such as nuclear power plants, and LLM-based hierarchical robotic planning with closed-loop self-correction. Across these works, the goal is to move safety beyond offline verification and embed it into the full autonomy stack, from high-level task planning to low-level motion control.

Selected Publications

  1. Online Control Barrier Function Construction for Safety-Critical Motion Control of Manipulators
  2. Safety-Aware Optimal Control for Motion Planning of Inspection Robots in Nuclear Power Plants Using Backward Receding SCA
  3. HiCRISP: An LLM-Based Hierarchical Closed-Loop Robotic Intelligent Self-Correction Planner

Power Systems

My work on power systems explores how learning-based controllers can improve stability and robustness in large-scale energy networks. In particular, I study model-free power system stability enhancement using dissipativity-based neural control, where neural controllers are trained with system-theoretic certificates rather than purely empirical objectives. This direction connects learning, dissipativity theory, and safety-critical control, with the aim of developing adaptive controllers that can enhance grid stability while retaining certifiable closed-loop behavior.

Selected Publications

  1. Model-Free Power System Stability Enhancement with Dissipativity-Based Neural Control

Large-Scale Scientific Infrastructure

In collaboration with Diamond Light Source, I studied electron beam stabilization for next-generation synchrotron operation, where fast orbit feedback must coordinate hundreds of sensors and actuators under bandwidth and actuator limitations. The proposed mid-ranging control method redesigns internal-model-based actuator coordination for systems with zero direct-current-gain actuators, derives closed-loop stability conditions, and validates the method on the Diamond-II beam stabilization problem. This work reflects a broader goal of bringing rigorous control design to high-performance autonomous infrastructure, where safety, reliability, and implementability are all essential.