• Real-time SLAM and navigation using Nav2
• Task-level control through finite state machines
• Integration with embedded systems (Arduino)
• Designed as a flexible playground for rapid robotics prototyping

Research Direction
ANIMA (Affective Natural Interaction through Memorable Animation) explores how animation-inspired expressive interfaces can enhance social presence, affective legibility and memorable interaction in humanoid robots. The project focuses especially on ocular expression as a core channel for non-verbal communication, while considering body gesture as a complementary and scalable expressive dimension.

• Parametric ocular expression model for gaze, blinking, eyelid aperture, pupil dynamics, asymmetry and expressive timing
• Animation-inspired design approach that prioritizes expressive legibility over anatomical facial realism
• Real-time simulation and monitoring interface for designing, testing and debugging robot expressive states
• ROS 2-oriented modular architecture for manual control, puppet control, perception inputs and future cognitive controllers
• Designed as an expressive HRI layer, currently under development using the Unitree G1 as a case study and reference platform

• Extended ROS 2-based navigation pipeline integrating SLAM and Nav2 for humanoid locomotion
• Designed MuJoCo → Gazebo synchronization pipeline for consistent state replication
• Converted Gazebo into a real-time visualization layer (“puppet mode”) driven by MuJoCo
• Resolved joint state desynchronization and timing issues affecting TF and control loops
• Implemented joint state bridge with timestamp correction for ROS 2 compatibility
• Optimized LiDAR configuration for stable SLAM and navigation
• Designed deterministic launch orchestration for full system startup (Gazebo, SLAM, Nav2)
• Addressed sim-to-real inconsistencies in collision modeling and physics behavior
• Mitigated odometry drift caused by impact noise during dynamic locomotion