For the autonomous stair climbing project on the Unitree Go1, I focused on enhancing the robot’s perception capabilities. I installed a Livox Mid-360 LiDAR and updated the URDF model to improve sensing, then integrated FAST-LIO SLAM into a C++ Kalman filter for state estimation within an OCS2-based NMPC-WBC framework. I also updated the robot-centric elevation mapping ROS package to leverage more accurate odometry, reducing drift and providing reliable terrain information for navigation. While my teammates concentrated on footstep planning and contact estimation, my work ensured the robot could perceive and interpret its environment accurately for safe autonomous climbing.