I am involved in a collaboration with the CNES in Toulouse on scheduling the scientific experiments of the space probe Rosetta/Philae on the comet 67P/Churyumov-Gerasimenko.
We have published a paper at CP 2012 and an extension in Constraints, as well as several vulgarisation articles were published: one in the magazine of the OR Society (Impact), one as an ACP's sucess story and on in the bulletin de la Roadef (in French).
Below is a description of the mission and of our role, taken from a paper currently submitted to Constraints.
Following the fly-by of the comets Halley and Grigg-Skjellerup by the spacecraft Giotto, an even more ambitious mission, including the landing of a robotic module on the comet nucleus, was approved by European Space Agency in 1993. This project involves more than 50 contractors from 14 European countries, Canada and the United States for developing the instruments necessary to a deeper study of the comet. The Rosetta spacecraft, embarking these scientific instruments, was then launched in 2004 by Ariane 5, and was set to travel more than six billion kilometers to finally reach the comet 67P/Churyumov-Gerasimenko in November 2014. Its complex trajectory includes four gravity assist maneuvers (three times around Earth and once around Mars) before finally reaching the comet and enter its orbit. During its travel, the probe has met two asteroids (Steins and Lutetia), and collected data and pictures. Upon arrival at 67P, Rosetta will enter orbit around the comet and follow it on its journey towards the Sun.
Philae features ten instruments, each developed by a European laboratory, to accomplish a given scientific experiment when approaching, or once landed on the comet. The exploratory mission will have three phases. First, the SDL (Separation-Descent-Landing) will run for 30 minutes during which many experiments will be done. Second, the FSS (First Science Sequence) will last 5 days. This phase is critical because the execution of the most energetically greedy experiments requires battery power. The quality of the schedule conditions the longevity of the batteries and is therefore a key to the success of the mission. Finally, during the LTS phase (Long Term Science), scientific tasks will be resumed at a much slower pace, using the lander's own solar panels to partially reload the batteries. This phase will continue for months until the probe is destroyed due to the extreme temperatures of the Sun.
The plans for each phase are elaborated on the ground at the Science Operations and Navigation Centre (SONC) in Toulouse. The problem is modeled as a constraint program. A software (called MOST) has been developed on top of the Ilog-Scheduler/Solver library by an industrial subcontractor to solve this constraint program.
In particular, constraints on the instruments energy usage, on data collection and transfers, as well as incompatibilities between experiments, have been implemented in this framework. Solutions are generated using classic depth-first search. Once they have been validated, these schedules are sent to the mission center located in Germany where they are encoded in the format expected by the Command and Data Management Systems (CDMS) onboard Rosetta and Philae. The communication delay to reach the spacecraft is roughly 25 minutes when approaching the comet. The CDMS can then execute the timed plan by triggering the different subsystems at the specified dates. It is important to mention that many parameters such as the orbital trajectory of Rosetta, which conditions the communication links with the lander Philae, will be known precisely only shortly before the landing. Currently, many different scenarios are solved and validated on the ground. However, the final and real plan will only be computed on yet partially unknown data, just a few days before being executed.