Speaker
Mr
Marco Ciarambino
(Politecnico di Milano - Aerospace Science and Technology Dept.)
Description
In recent years, a renewed interest in space exploration has induced investing a growing amount of human and financial resources to provide next generation spacecraft with enhanced autonomous navigation and landing capabilities. Complex missions in which close approach to and landing on uncooperative objects play a major role are being developed by numerous space agencies: in particular, ESA is working together with ROSCOSMOS on a cooperative programme for Mars, Phobos and Moon exploration: as our satellite is concerned Luna-Resurs Lander (Luna-27) mission planned for 2020 and the Luna 29, the Lunar Sample Return mission to follow are involved, strongly focused on the South Pole landing to collect icy volatiles located in a very precise region of the huge Aitken crater. Part of the European contribution for the Luna-27 mission is the PILOT (Precise and Intelligent Landing using Onboard Technologies) subsystem, for enhancing autonomous landing capabilities in terms of high precision landing and hazard detection and avoidance functionalities. An analogous collaboration has been established for the ExoMars programme, which include a lander delivery first, followed by a rover release on Mars, to be launched in 2016 and 2018 respectively and for the exploration of the small Mars moon with the Phootprint Mission, a Phobos Sample Return mission planned for the Twenties.
Since 2006, technologies for autonomous landing are studied by NASA in the frame of the Autonomous Landing Hazard Avoidance Technology (ALHAT) program. ALHAT technologies, tested at the end of 2014 in free flight on the Morpheus lander demonstrator, are going to be integrated in future exploration missions. Recently CNSA has performed its first lunar landing with a lander/rover system with the Chang'e 3 mission (with missions 4 and 5 already scheduled), while ISRO is planning to put a lander carrying a rover on the lunar surface by the early 2018 in the Chandrayaan 2 mission. Among the various technologies under study, vision-based systems represent one of the most promising tools to provide the required level of accuracy, unattainable by classical technologies. In this paper the research carried out at the Department of Aerospace Science and Technology (DAER) of Politecnico di Milano about algorithms and tools dedicated to autonomous navigation and hazard detection is presented.
A hazard detection algorithm, based on artificial neural networks, has been developed and extensively tested. A single grayscale image of the landing area is filtered to extract essential information regarding shadows, surface roughness and slopes, and then it is analyzed by a cascade neural network, which provides a hazard map. Hence, hazard information is exploited by a subroutine that computes the most suitable landing site, taking into account safety requirements and trajectory constraints. The achieved computational efficiency allows the system to operate real time.
Also a vision-based relative navigation tool, relying on features tracking between subsequent frames, is currently under development. Camera information is fused together with measures coming from classical sensors, like Inertial Measurement Units and radar/laser altimeters. Collected data are filtered, taking into account the lander dynamics to obtain a convergent estimate of the system status. Collected information are also exploited to build a dynamical semi-dense map of the landing area, exploitable by hazard detection to estimate slopes and to locate the selected target.
To further increase the TRL of the aforementioned algorithms, an experimental facility is under setup at DAER premises. Since the scarce availability of complete real landing imagery datasets, vision-based algorithms development relies widely on synthetic images. To validate such approach, experiments are necessary. Moreover, the whole navigation system performance can be assessed only connecting the composing parts together, to verify mutual influences. The experimental facility under development represents a Hardware-in-the-Loop environment, and it is composed by a 3D mock-up simulating the planetary surface; a 7 DoF robotic arm to carry the sensors suite and simulate the spacecraft dynamics; an illumination system consisting of a LED array and a dimming subsystem to provide a realistic and controllable light. Hence the aim is to simulate the landing maneuver with the robotic arm in a scaled environment with realistic illumination conditions, over a reproduced planetary surface. The system is designed to verify either hardware and software breadboards up to TRL 4, with possible further enhancements to qualify flight models to TRL 5. The development status of the vision-based tools is presented. The design and the first activities for functional verification of facility components are shown, such as first tests of the hazard detector in a lunar environment.
- B. Houdou et al., "Roscosmos-ESA Cooperation in Lunar Exploration", 65th International Astronautical Congress - IAC 2014, 29 September - 3 October, Toronto, Ontario, Canada.
- P. Lunghi, M. Ciarambino, M. Lavagna, "A Multilayer Perceptron Hazard Detector for Vision-Based Autonomous Planetary Landing", AAS/AIAA Astrodynamics Specialist Conference 2015, 09-13 August, Vail, CO.
- P. Lunghi, M. ciarambino, M. Lavagna, "A New Test Facility for Vision-Based Hazard Detection and Avoidance Systems for Planetary Landing Maneuvers", 66th International Astronautical Congress - IAC 2015, 12-16 October, Jerusalem, Israel.
- P. Lunghi, M. Lavagna, R. Armellin, "A Semi-Analytical Guidance Algorithm for Autonomous Landing", Advances in Space Research, Vol. 55, No. 11, 2015, pp. 2719-2738.
| Applicant type | First author |
|---|
Primary authors
Mr
Marco Ciarambino
(Politecnico di Milano - Aerospace Science and Technology Dept.)
Mr
Paolo Lunghi
(Politecnico di Milano - Aerospace Science and Technology Dept.)
Co-authors
Mr
Luca Losi
(Politecnico di Milano - Aerospace Science and Technology Dept.)
Prof.
Michelle Lavagna
(Politecnico di Milano)
