Year: 2018
Type: conference-paper
Abstract: In all orbital applications, such as on-orbit servicing and repair, rendezvous and docking, active debrisremoval (ADR), and planetary applications, such as exploration of unknown environments for scientificpurposes by means of rovers, GPS-denied navigation aspects have a very large impact on the successfuloutcome of missions. Having a sensor suite, and hence several different sensors, also requires, at the sametime, a suite of navigation algorithms able to deal with different kinds of inputs. Some of them, however,can be shared between multiple sensors, after thorough pre-processing of the raw data. Additionally, thesame kind of sensor can require two different navigation algorithms depending on the scenario. The workdescribed in this paper aims to present and critically discuss the approach to precise relative navigationsolutions with a complete suite of sensors and their performance in different space-oriented applicationscenarios.Standalone navigation filters are examined. In the case of a high-resolution camera for an orbitalscenario, the pose of a target, with respect to a chaser, can be thoroughly obtained with the aid offiducial markers. Stereo camera-based navigation is also addressed with visual odometry. In the case of astereo camera the problem of scale estimation during odometry is solved by means of triangulation. Sincethe outputs of the sensor-suite are also dense 3D point clouds, Iterative Closest Point and Histogram ofDistances (HoD) with Kalman filter approaches are analyzed, paying attention to the provision of correctsensor characterization. The results for each filter are exhaustively examined, highlighting their strengthsand the points where some improvements can be achieved
Source: CERES (Cranfield University)
Language: English
Subfield: Aerospace Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
Funder: Cranfield University
OpenAlex ID: W2909395403