Multi-Robot Localization and Tracking

Undergraduate project at Hanyang Univ.

This project developed a centralized perception and coordination framework for multiple table robots in an indoor environment. For full project scope and team-level implementation details, see the project repository.

My Contribution

1. Localization

   • Computed ICP-based LiDAR odometry and fused IMU, wheel-encoder odometry, and LiDAR odometry in an onboard EKF estimator.
   • Estimated final robot states through confidence-weighted interpolation between robot-internal EKF estimates and ceiling-camera observations.

2. Vision system using ceiling-camera

   • Estimated robot position w.r.t camera coordinates utilizing depth measurements and YOLOv8 (for robot detection and tracking), providing external information for robust robot localization.
   • Identified indoor obstacles for ensuring safe and accurate path planning.

Outcome

• Validated the integrated system through a successful end-to-end demonstration in a real indoor environment.
• Established a practical localization stack combining onboard state estimation and infrastructure-based visual perception.
• Improved robustness of environment-aware multi-robot operation through joint robot and indoor-scene detection.