Stereovision-based pose and inertia estimation of unknown and uncooperative space objects

Spacecraft autonomy is an important aspect of space mission designs. The present study focuses on autonomous spacecraft proximity operations while paying attention to the estimation of pose, inertia and motion properties of uncooperative objects. Accurate estimation of pose and motion behavior of unknown objects, for instance, asteroids and resident space objects, is necessary for various applications.

Accurate motion estimation of unknown objects allows for docking, monitoring and autonomous inspection. However, working with an uncooperative space object is quite challenging owing to lack of information about the structure as well as the motion of the target. The interest of various space agencies has been the automation of space operations and missions because of its vast applications. This comes with the benefit of mitigating high costs and risks emanating from the presence of humans on board.

In a recent paper published in Advances in Space Research Professor Riccardo Bevilacqua at the University of Florida in collaboration with Vincenzo Pesce and Professor Michele Lavagna at Politecnico di Milano in Italy developed a novel approach to approximate the velocity, attitude, relative position and angular velocity and the ratios of the inertia matrix components of an uncooperative space object implementing stereo-vision measurements. The ratios of the inertia matrix were added to the state and a pseudo-measurement equation developed in the proposed observation model.

The authors intended to enhance the relative state estimation with respect to the present literature works, implement a novel method through the estimation of inertia matrix component ratios, and reconstruct the inertia matrix. They adopted an observation model implementing Iterated Extended and Extended Kalman Filters for the observation procedure.

The Kalman Filter method allows for solving discrete-data filtering problems in a computationally efficient way. The Extended Kalman Filter, which is a powerful estimation tool, was used in cases with behavior based on non-linear models. With the method, the authors were able to perform re-linearization and local iterations for the modified reference state. The method maintained its computational efficiency and simplicity throughout the iterative cycle.

For numerical analysis, the authors used a Monte-Carlo analysis in evaluating the performance of the filter. They considered a space object in low earth orbit and a satellite. The authors observed that in the filtering process, the inertia components converged without equality constraint but only with a higher target angular velocity.

The results obtained indicated some convergence of estimation errors for every considered quantity. There were some improvements when compared to previous work dealing with the same problem. The authors presented a video processing method for reconstruction of the body’s geometrical properties using cameras.

This paper developed a new algorithm that can be used to estimate the motion, pose as well as inertia properties of an uncooperative space body. The outcomes of the study indicate how the proposed algorithm allows for accurate estimation of the complete relative state as well as the inertia component ratios. The inertia components converge without equality constraint, however, with a higher target angular velocity. Above all, a video processing approach was proposed to reconstruct the geometrical attributes of the object using cameras.