This thesis describes the design and implementation of an algorithm for tracking a moving (e.g., `tumbling') target. No a priori information about the target is assumed, and only a single camera is used. The motivation is to enable autonomous rendezvous, inspection, and docking by robots in remote environments, such as space and underwater. Tracking refers to the simultaneous estimation of both the target's 6DOF pose and 3D shape (in the form of a point cloud of recognizable features), a problem of the SLAM (`Simultaneous Localization and Mapping') and SFM (`Structure from Motion') research fields. This research extends SLAM/SFM to deal with non-communicative moving targets (rigid bodies) with unknown, arbitrary 6DOF motion and no a priori knowledge of mass properties, dynamics, shape, or appearance. Specifically, a hybrid algorithm for real-time frame-to-frame pose estimation and shape reconstruction is presented. The algorithm combines concepts from two existing approaches to pose tracking, Bayesian estimation methods and nonlinear optimization techniques, to achieve a real-time capable, feasible, smooth estimate of the relative pose between a robotic platform and a moving target. The rationale for a hybrid approach is explained, and an algorithm is presented. A specific implementation using a modified Rao-Blackwellized particle filter is described and tested. Field demonstrations were performed in conjunction with the Monterey Bay Aquarium Research Institute, using the camera-equipped Remotely Operated Vehicle (ROV) Ventana to observe, reconstruct, and track the pose of an underwater tethered target in Monterey Bay. Results are included which demonstrate the performance and viability of the hybrid approach.