Controlling spacecraft

As exploratory space missions take unmanned autonomous scientific craft further and further from Earth, robust spacecraft control systems are increasingly vital.

This is why an international team of engineers is using statistical and modelling techniques to improve fault tolerance and recovery to help ensure the success of ESA missions.

Dr Declan Bates, senior lecturer in control engineering, will lead a Leicester University group, part of a consortium investigating the issue, which also includes Spanish technology company GMV, Canada's NGC Aerospace and Oxford University.

The Leicester engineers were invited to join the project because of their expertise in using analysis techniques to make control systems resilient.

'We're trying to verify correct functioning of the control systems in the face of various kinds of uncertainty in the control systems and in the environment in which they operate,' said Bates. 'In recent years, we have been more interested in looking at space applications, so we already had a well-established reputation in this area.'

Bates explained that the key challenges of this type of space mission stem from the fact that the systems have to be autonomous from start to finish.

One key scenario the project will address is the control of a group of satellites, such as those used to investigate the surface of Mars prior to a manned mission.

'When groups of satellites have to rendez-vous to form a single satellite or to form a set position relative to each other in space, you want to make sure this capability is preserved even in the face of uncertainties in their model of the situation,' said Bates. 'To ensure the rendezvous happens correctly and handles any faults or failures, we run simulations beforehand of problems that might arise in different parts of the system.'

Bates explained the standard way to ensure fault tolerance is to run simulations based on Monte Carlo statistics — where thousands of simulations are run covering different types of uncertainty and disturbances to ensure that statistically it is very unlikely anything will cause a problem.

'The problem with this is that it's complicated and very time intensive,' said Bates. 'So what we propose to do is develop more analytical techniques to prove that X or Y cannot occur for this level of uncertainty without needing to run so many simulations.'

Optimisation techniques are planned to be used by the team to focus on finding the very worst case scenarios. 'You can try to identify deterministically what's the very worst thing that can happen for a given level of uncertainty and disturbances and ensure that worst thing can be handled,' said Bates.

The team liaises closely with ESA engineers who are deeply involved at a technical level to determine what situations are likely to occur and define the problems. They will analyse the results of the scenarios and aim to improve the results.

Simulation-based scenarios will be developed by the consortium involving models of satellites, mission trajectories and requirements they can simulate on a computer. They will then draw up a list of situations and analyse the performance of the control systems as they are being used. Companies involved have access to near mission-identical simulation platforms.

As there will always be a slight difference between the real system and a model, the researchers will also quantify this error and include it in the simulation to make sure that the controller will function on the real mission and not just on the computer model.

Though the key aim of the project is to correct faults before launch, certain parts of the control systems will have an autonomous capability to reconfigure themselves if faults arise. Those systems are based on models of various types of environmental disturbance, or problems that might occur in space.

Leicester's main contribution will be designing and analysing the algorithms implemented in the control systems and trying to evaluate their robustness.

Oxford will be more focused on developing theoretical analytical techniques. GMV and NGC will develop the examples to demonstrate the techniques, implement them and compare with current procedures.

The project will run for around two years. It will result in new analytical techniques that ESA can apply to any future space missions.

Berenice Baker