Indoor Person-Following Stack (ROS + Jetson)
Fused LiDAR and RGB-D for robust indoor tracking with real-time constraints.
Outcome. Stabilized person-following in clutter by fusing LiDAR + RGB-D and profiling end-to-end latency on Jetson.
Role: Intern researcher · Dates: Dec 2022 – Feb 2023 · Stack: ROS, C++/Python, Deep SORT, Jetson
Highlights
- Tuned Deep SORT and data association to cut ID-switches; analyzed latency budgets per perception stage.
- Implemented recovery behaviors (re-acquire, obstacle bypass) validated in field tests.