Applications of Computer Science to Space Applications

COMP036

Tuesday June 12, 2007
Institut d’Informatique (Séminaire 3), FUNDP, Namur

UCL and FUNDP are organizing two MoVES Doctoral School lectures by international experts in the application of two different computer science disciplines, namely verification and planning/scheduling, to ESA and NASA space applications.

Program

Software Verification for Space Applications
Dr. Guillaume Brat, RIACS / NASA Ames, USA

9h00-10h30: Part I (slides)

In this lecture, I will describe how we applied static analysis to the verification of flight software at NASA. I will explain why disappointing experiences with commercial tools (lack of scalability and precision) led us to develop our own static analyzer, C Global Surveyor (CGS), and, will go on with a detailed description of the tool. CGS has been developed following the abstract interpretation framework, and, it has a few interesting features such as being able to distribute the analysis over several processors or performing dynamic clustering of variables during abstraction. I will conclude by taking a look at our future directions in static analysis, including coupling CGS, or its successor, with a symbolic execution engine.

11h00-12h30: Part II (slides)

In this lecture, I discuss how we approach verification and validation for ground software, i.e., software used on the ground to monitor and control spacecrafts. I will discuss the issues of applying formal methods to the verification of human-machine interfaces. I will describe the different models involved (machine, display, and users) and how we use them to validate interfaces. I will also describe our current efforts in the verification of planning and executing systems. These systems are expected to provide substantial improvements in terms of adaptability, response time, and reliability for ground software.

Planning and Scheduling for Space Applications
Ir. Marc Niézette and Dr. Sylvain Damiani, Vega Group PLC, Germany

14h00-15h30: Part I (slides)

In the space domain, mission planning covers the generation of timelines of activities to be carried out by the various entities of a space mission in order to achieve its goal. This lecture will provide a survey of on-ground planning systems operated in missions of the European Space Agency. We will introduce the requirements on such systems, starting with situating the mission planning functionality in a typical space mission ground segment, and introducing the characteristics of the typical planning problem(s) in several contexts (earth-observation, deep-space, navigation, ground station network). The mission planning software architecture developed by VEGA will be described, together with the rationale for our approach to mission planning systems development. We will conclude by looking at the recent developments in automated planning and autonomy in ESA and in Europe.

16h00-17h30: Part II (slides)

In this lecture, we will present in more detail the techniques and algorithms underlying the implementation of two specific systems. We will start with the ESA ground station network planning, an operational system that is providing the automated allocation of ground stations services to all spacecraft managed by ESA. The modelling of mission requirements and network constituent capabilities will be introduced. We will show how a constraint network can be constructed from the dynamic input to the system and its configuration, and present the algorithm used for the resolution of the constraint network. The second part of the lecture will introduce the Automated Planning System (APS) for science operations, which was demonstrated on the ESA Venus-Express mission, and is currently being evaluated for its applicability in the context of future missions of the ESA deep-space programme. The specificity of this planning domain requires interfacing with complex simulators and models. The system implements goal-oriented planning, with optimization of the science return from the plan.

About the Speakers

Dr. Guillaume Brat received his M.Sc. and Ph.D. in Electrical & Computer Engineering in 1998 (The University of Texas at Austin, USA). His thesis defined a (max,+) algebra to model and evaluate non-stationary, periodic timed discrete event systems. Since then, he has specialized on the application of static analysis to software verification. From 1997 to June 1999, he worked at MCC where he led a project that developed static analysis tools for software verification. In June 1999, he joined the Automated Software Engineering group at the NASA Ames Research Center and focused on the application of static analysis to the verification of large software systems. For example, he co-developed and applied static analysis tools based on abstract interpretation to the verification of software for the Mars PathFinder, Deep Space One, and Mars Exploration Rover missions at JPL, various International Space Station controllers at MSFC, and the International Space Station Biological Research Project at the NASA Ames Research Center. His current interests also include the verification of planning and execution systems.

Marc Niézette graduated in 1988 as Electrical Engineer from the University of Liège (Belgium). He joined the VEGA Group PLC in 1997, where he is leading the Mission Planning and Payload Data Systems Group in Darmstadt, Germany. He has been working in the field of Mission Planning Systems (MPS) for the last 12 years, during which time he has led a number of ESA planning systems developments and studies as VEGA Project Manager.

Dr. Sylvain Damiani graduated in 2002 as Space Systems engineer from Supaero, Toulouse, France, which was followed by a Ph.D. in Artificial Intelligence carried out at ONERA (French Aerospace research organism) for CNES (French Space Agency) in Toulouse. His thesis focused on autonomous decision requirements for space systems, in particular on-board real-time decision making. In the context of the management of an Earth watching constellation of satellites, it included the definition and the implementation of planning algorithms (for observation and data downlink activities), and the simulation of the complete space system using a multi-agent framework. Since then, he has been working at VEGA in the Mission Planning and Payload Data Systems group, where his tasks have included the design and the implementation of planning algorithms.

Organization

The lectures are organized as part of the MOVES Doctoral School and as a satellite event of AFADL'07. They will be given in English. The lectures will take place in room Séminaire 3 (3rd floor) of the Institut d’Informatique at FUNDP. Attendance is free to graduate students and researchers from all universities. Participants should register by sending an e-mail to Charles Pecheur.

Contact

courses/bratniezette.txt · Last modified: 2007/06/19 11:54 by charles.pecheur