Methods and systems for object tracking are disclosed in which the bandwidth of a "slow" tracking system (e.g., an optical tracking system) is augmented with sensor data generated by a "fast" tracking system (e.g., an inertial tracking system). The tracking data generated by the respective systems can be used to estimate and/or predict a position, velocity, and orientation of a tracked object that can be updated at the sample rate of the "fast" tracking system. The methods and systems disclosed herein generally involve an estimation algorithm that operates on raw sensor data (e.g., two-dimensional pixel coordinates in a captured image) as opposed to first processing and/or calculating object position and orientation using a triangulation or "back projection" algorithm.
