I am currently engaged in the AI-Inertial project, which leverages inertial sensor data to enhance indoor-outdoor positioning systems. Our landmark detection methodology utilizes magnetic identification, achieving an accuracy rate of approximately 80% on our proprietary datasets and around 70% on third-party datasets. Additionally, I am refining the inertial-based Pedestrian Dead Reckoning (PDR) to mitigate estimation errors inherent in traditional PDR methods.Previously, my research focused on the In-Out Door Location System, integrating outdoor GNSS data with indoor radio maps and floor plans. For backend radio map collection, I employed robotics platforms, including TurtleBot2 and Spot Micro, in conjunction with Visual Simultaneous Localization and Mapping (VSLAM) to generate comprehensive localization maps. My work also involved Reinforcement Learning to develop Complete Coverage Path Planning (CCPP) in 3D for robotic automation.On the frontend, I am designing algorithms to deliver robust and stable localization results, which are subsequently implemented into Application Specific Integrated Circuits (ASIC) for potential end-user device integration. Our efforts also include developing noise-canceling methods for IMU sensors and enhancing Pedestrian Dead Reckoning (PDR).My primary programming languages are Python, C, and C++, in which I possess professional-level proficiency. I also have foundational knowledge in several other programming languages from my bachelor's degree studies and am keen to further hone my skills across diverse fields.