Improving Multi-Target Tracking for Gimballed Camera Systems

Dr. Bethany Allik, Senior Research Scientist at Trillium Engineering, recently presented a conference paper at the AIAA SciTech Forum, the world’s largest aerospace event, addressing a key challenge in aerial ISR: maintaining reliable multi-target tracking when using gimballed cameras with limited fields of view and imperfect attitude information. She also co-chaired the technical session, helping lead discussions on navigation, estimation, sensing, and tracking.

Addressing Real-World Sensor Constraints

Successfully tracking multiple ground objects from aerial gimbaled camera systems introduces unique challenges that traditional tracking methods do not always handle well. For example, when targets move outside the camera’s field of view, their tracks are often prematurely terminated. This makes it harder to re-acquire those targets later

This paper presents a novel multi-target tracking algorithm that accommodates noisy camera attitude and utilizes the Poisson Multi-Bernoulli Mixture (PMBM) filter.

A New Multi-Target Tracker Specifically Designed for Gimballed Cameras

The Trillium solution introduces an innovative filter update strategy designed specifically for aerial gimbaled camera systems. This allows the tracker to successfully manage tracks that leave and re-enter the field of view during the mission.

Enabling Track Revisit and Preserving Continuity

By modeling and maintaining tracks on objects that are temporarily unobservable, the Trillium solution supports revisit and re-acquisition of objects, improving continuity and robustness in dynamic operations.

Read the full white paper for a detailed technical breakdown of the MTT formulation, update strategy, and results.