Scholarlink

In this study, we aim to improve the rotational motion and inertia parameter estimation performance of the Unscented Kalman Filter (UKF) for a torque-free tumbling non-cooperative space object using the Gaussian Process (GP) under long-term occlusion. Occlusion is a quite common scenario for space missions and the UKF model cannot estimate the motion parameters with high accuracy if occlusion persists for a long time. Hence, we generate GP models to predict the sensor measurements during the occlusion and utilize the predicted measurements to update the UKF-GP fusion model. This study assumes that the measurement sensor is a stereo-camera system that collects the 2D projections of the positions of five features on the non-cooperative space object. Results show that the UKF-GP fusion approach which is inspired by the Dynamic Data Driven Applications Systems (DDDAS) concept leads to convergence even during occlusion, which the traditional UKF is unable to do.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Improving Physics-Based Motion and Physical Parameter Estimations of a Tumbling, Non-cooperative Space Object Through DDDAS

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Rabiul Hasan Kabir,
  • Xiaoli Bai

摘要

In this study, we aim to improve the rotational motion and inertia parameter estimation performance of the Unscented Kalman Filter (UKF) for a torque-free tumbling non-cooperative space object using the Gaussian Process (GP) under long-term occlusion. Occlusion is a quite common scenario for space missions and the UKF model cannot estimate the motion parameters with high accuracy if occlusion persists for a long time. Hence, we generate GP models to predict the sensor measurements during the occlusion and utilize the predicted measurements to update the UKF-GP fusion model. This study assumes that the measurement sensor is a stereo-camera system that collects the 2D projections of the positions of five features on the non-cooperative space object. Results show that the UKF-GP fusion approach which is inspired by the Dynamic Data Driven Applications Systems (DDDAS) concept leads to convergence even during occlusion, which the traditional UKF is unable to do.