Investigation on space navigation techniques based on optical sensors

Space navigation deals with the determination of the kinematic state (position, velocity, attitude) of a spacecraft. The kinematic state can be obtained based on the output of suitable sensors and by means of appropriate computation. As far as it concerns the sensors, the optoelectronic ones – already present – are facing increasing interest and applications. The success of these sensors depends on the improved performance, the reduced cost (also as a result of a strong commercial growth for parent terrestrial products) as well as on the availability of computation resources required to efficiently process the images. It clearly appears that the interest for this sensing technology will continue in the foreseeable future, and their application will spread to all classes of space platforms, including the small cubesats. The thesis is devoted to investigate some aspects related to the use of the optoelectronic sensors – indeed the word techniques – on-board spacecraft. First, the focus is on the star tracker, deemed as the most accurate (and the most expensive) of the attitude sensors, and considered to be the flagship of the space optoelectronic instruments, with complex hardware and strong computational requirements. Star trackers are optoelectronic instruments providing the attitude of the satellites through star observations. The estimate of spacecraft attitude is obtained beginning with the measurements of star coordinates in the body reference frame and comparing these “observed” coordinates with the “known” star directions stored in the on-board star catalogue. A lot of studies have been done in order to improve the attitude estimate accuracy and to define faster algorithms capable to be implemented on-board of satellite. Three different methods will be presented for following applications: TRIAD, q-method and QUEST. Indeed, the two initial chapters are devoted to resume star tracker basics and to recall the attitude determination techniques. Then, the following part of the thesis (chapter 3) deals with some more original contribution considering the calibration of the star trackers. This is a topic of high current interest as it greatly affects the cost of the hardware. Instead of carrying on a long expensive test campaign at the facility, a simpler and faster two steps process with a “raw”, preparatory phase at the production site and a final, possibly autonomous, accurate calibration once in orbit can produce valuable results. In such a way it is possible to reduce the effort at the factory and to directly evaluate the performance once in orbit. Indeed, small deviations in the equipment occurring during the most critical condition, i.e. the launch phase, can be still corrected before real measurements campaign will begin. Clearly this approach becomes extremely interesting for the new instruments built in large batches for huge Low Earth Orbit formations’ satellites. The third chapter is focused on the calibration process (on-ground and on-orbit calibration) and reports the simulations and the findings for this proposed technique. Some more general discussion is required to introduce the last part of this dissertation (chapter 4). Space probes increasingly explore the solar system, up to faraway planets. Orbit determination of these probes, based on radio tracking from Earth, becomes clearly less accurate as the distance from Earth grows up. Above all, the time required for telemetry/navigation data downlink and tele-command uplink also increases with distance from Earth and therefore real-time manoeuvres and operations become impossible. As an example, the time needed to send a telecommand or receive telemetries in the Rosetta mission was about 20 minutes once the probe reached the target. When a spacecraft is close to a planetary target (or celestial body, including comets and asteroids), optical navigation – in use since the experiments with Mariner 6 and 7 missions to Mars (1969) – can nowadays ensures accurate estimates of the relative kinematics and allows to conceive manoeuvres computed on-board, autonomously and in real time. This technique, based on imaging and on the comparison with already known data as previously captured images, celestial catalogues or ephemerides, helps with the determination of the complete kinematic state of the spacecraft, relative to the target. Indeed, it is similar to attitude determination traditionally carried out by means of star trackers, where the spacecraft’s orientation is computed thanks to a priori information included in the star catalogue. The similarity in concept, with imaging process and comparison to stored information, introduces the question if star tracker’s and proximity cameras’ functions can be exploited by the same on-board hardware. The availability of a universal optical navigation sensor, sharing a large part of its expensive components, could really be an enabling technology for a more effective space exploration. The aim of this part of the work is to investigate and analyse the possibility of such a universal sensor, which is collecting more and more interest. The main issue is the identification of the sensor’s configuration – as an example beginning with multi-head star trackers with different optics and focal lengths – and algorithms allowing to improving star trackers performances and to exploit this twin use. This identification moves through a correct modelling of the sensor behaviour. The combination between star trackers and proximity cameras as position/attitude sensors could obviously allow a reduction in costs, and – probably more important at the current, preliminary status of this approach – provide a back-up solution in case of failures thanks a possible, even non-optimal redundancy. Furthermore, the interest of this study is not limited to deep space missions, and may be extended to other vehicles currently using star trackers and cameras as the planetary rovers. In the first part of chapter four will present a typical optical navigation system and the method used for the estimates of kinematics parameters. Then the discussion will be focused on the use of the star tracker as backup or in place of the navigation camera during the main phases of the mission: cruise, approach and fly-by or descent to the target. A simple case study, relevant to a low altitude lunar orbit, will be reported and its results will be presented and discussed. For that simulation, the star tracker is able to compute the position of the spacecraft with respect to the planet inertial reference frame using the landmarks catalogue, as for the estimate of attitude. The capability of a multi-head star tracker to estimate the relative position of the spacecraft with respect to a target, therefore acting as a navigation camera, opens the path to a universal optical sensor. The use of this sensor will be for sure limited to specific mission phases, due to the lenses’ limitations and to the threshold associated to the detector. As an example, the approach to deep space celestial bodies (asteroids, far planets) can be considered as a possible application regime. At least in these specific phases, the proposed solution has the potential to reduce the costs and/or offer a redundancy in case of failure of part of the instruments. Indeed, the analysis of this extended application of the star tracker is quite interesting for future deep space missions. Furthermore, the interest of the study is not limited to interplanetary navigation, and can be extended – by means of using multiple heads or specific filters - to other vehicles currently using star trackers and cameras as the planetary rovers.