Collaborative SLAM with Moving Object Tracking

Dynamic environments present significant challenges for autonomous robots because moving objects violate the assumptions made by conventional SLAM algorithms. This project developed a unified Bayesian framework that simultaneously estimates robot trajectories, static environmental features, and moving objects.

Unlike traditional approaches, the algorithm automatically distinguishes between static and dynamic features using Random Finite Set theory without requiring prior knowledge of object identities.

Research Contributions

• Unified collaborative SLAM and moving object tracking.
• Automatic separation of static and dynamic features.
• Bayesian estimation using Random Finite Set theory.
• Validation using multiple mobile robots.

Demonstration Video

Publications

D. Moratuwage, B. Vo and D. Wang,

“Collaborative Multi-vehicle SLAM with Moving Object Tracking,”

IEEE International Conference on Robotics and Automation (ICRA), 2013.

Future Directions

This work established the foundations for probabilistic perception in dynamic multi-robot environments.