Model Based Fault Detection and Diagnosis for Planetary Exploration Rovers Using Inverse Simulation

Brief description of the context of the research including potential impact:
The focus of this project is the development of a health monitoring and fault recovery architecture that will be used in the context of planetary rover guidance, navigation and control. This architecture will incorporate both forward and inverse simulation techniques, fault and signal modelling processes and a range of different fault detection techniques such as limit-checking, process-identification and parity equations. The performance of the proposed architecture will be compared against a selection of existing detection and isolation methods. Analysis of this comparison will provide an outline the benefits and disadvantages of using this approach in the context of a planetary rover. A number of faults of varying complexity typical to those experienced by a rover's systems will be considered in this study. In addition, the effect of terra-mechanics on the performance of the rover and the health monitoring and fault recovery system will be investigated.
The development of this architecture will impact on a wide range of engineering applications where health monitoring and recovery from systemic faults would extend operational lifespan. In particular this architecture would be extremely useful for autonomous vehicles (e.g. rover, drones, deep space probes) where the operational environment is too remote for human intervention when a fault occurs. In addition, it would be very useful for chemical processes and manufacturing systems for accurately monitoring and identifying the location of faults.

Aims and objectives:

The development of fault detection and isolation algorithms using forward and inverse simulation techniques
The development of a simulation environment that accurately represents the dynamics of a planetary rover and the terra-mechanics experienced on a planet's surface.
The implementation of the fault detection and isolation architecture to the health monitoring of typical rover systems e.g. sensors and actuators.
Compare the performance of the proposed architecture against other conventional fault detection and isolation methods.
To develop and implement appropriate recovery procedures in response to typical faults that could be experienced the key systems within a planetary rover.

Novelty of the research methodology:

The study and implementation of Fault Detection, Isolation and Recovery is a growing area of research, particularly for application to autonomous systems and vehicles. The use of Inverse Simulation for fault analysis is a novel area for health monitoring research. Its application in the field of planetary exploration rovers is equally novel. Naturally the proposed comparison between this architecture and more conventional methods would be another novel part of the study.

Alignment to Research Council's strategies and research areas:

Although this project is connected to a number of the EPSRC's research areas (e.g. Artificial Intelligence Technologies, Control Engineering, Sensors and Instrumentation, Human-Computer Interaction), it is directly aligned with the EPSRC's Robotics research area. The intended outcomes from this study are aligned with the industrial focussed strategy for this area that has identified the need for research that creates new capabilities in extreme environments and ensures safe/efficient vehicle and manufacturing systems.

Any companies or collaborators involved:

Although not companies or collaborators are directly involved with this project, it stems from work funded by the UK Space Agency which involved direct collaboration with Airbus and ESA.