Robotics · Reinforcement learning · Human-robot interaction
PhD candidate at the University of Louisville
Member of Telos Lab
I research reinforcement learning for embodied robots, with a focus on legged locomotion, sim-to-real transfer, and control that remains useful beyond a single training environment.
My work also explores tactile sensing and physical human-robot interaction—connecting learning algorithms with the practical realities of sensing, evaluation, and deployment.
Research record
Journal (PREPRINT) · 2026
Preprints.org · Version 1
A dual-loop neuroadaptive tactile interface was evaluated across 28 participants and 112 personalized calibration models, enabling real-time human-guided robot-arm motion.
Accepted conference paper · 2026
IEEE National Aerospace and Electronics Conference (NAECON) · Accepted
A token-compressed temporal transformer fuses proprioception with a low-cost 64-point time-of-flight terrain scan and transfers to a physical ROBOTIS OP3 for terrain-aware locomotion.
Accepted conference paper · 2026
IEEE National Aerospace and Electronics Conference (NAECON) · Accepted · † Equal contribution
A virtual-reality teleoperation pipeline in Isaac Sim and Isaac Lab collects reusable cube-stacking demonstrations, retargets them to joint space, re-renders observations, and expands the dataset with Isaac Mimic.
Conference paper · 2026
2026 IEEE/SICE International Symposium on System Integration (SII) · pp. 1259–1264
A torque-control reinforcement-learning framework for the Unitree B1 traverses rough terrain, tracks commanded velocity, and climbs stairs in Isaac simulation without exteroceptive sensing.
Conference paper · 2025
2025 IEEE-RAS 24th International Conference on Humanoid Robots (Humanoids) · pp. 1233–1240
A full-body telepresence stack for the ROBOTIS OP3 combines virtual-reality upper-body control with reinforcement-learning locomotion for compliant tele-loco-manipulation.
Conference paper · 2025
2025 IEEE 21st International Conference on Automation Science and Engineering (CASE) · pp. 2194–2199
A reinforcement-learning locomotion pipeline trains in NVIDIA Isaac Sim and transfers to a physical Unitree Go1 for real-world evaluation.
M.S. thesis · 2024
Electrical Engineering · University of Louisville
OctoCan tactile sensing and neuroadaptive control support physical robot guidance, while online dynamic time-warping enables speed-adaptive imitation with the Zeno humanoid.
