Reconfigurable autonomy for future planetary rovers.

Extra-terrestrial Planetary rover systems are uniquely remote, placing \nconstraints in regard to communication, environmental uncertainty, and \nlimited physical resources, and requiring a high level of fault tolerance \nand resistance to hardware degradation. \nThis thesis presents a novel self-reconfiguring autonomous software architecture \ndesigned to meet the needs of extraterrestrial planetary environments. \nAt runtime it can safely reconfigure low-level control systems, \nhigh-level decisional autonomy systems, and managed software \narchitecture. The architecture can perform automatic Verification and \nValidation of self-reconfiguration at run-time, and enables a system to be \nself-optimising, self-protecting, and self-healing. A novel self-monitoring \nsystem, which is non-invasive, efficient, tunable, and autonomously deploying, \nis also presented. \nThe architecture was validated through the use-case of a highly autonomous \nextra-terrestrial planetary exploration rover. Three major \nforms of reconfiguration were demonstrated and tested: first, high level \nadjustment of system internal architecture and goal; second, software \nmodule modification; and third, low level alteration of hardware control \nin response to degradation of hardware and environmental change. The \narchitecture was demonstrated to be robust and effective in a Mars sample \nreturn mission use-case testing the operational aspects of a novel, \nreconfigurable guidance, navigation, and control system for a planetary \nrover, all operating in concert through a scenario that required reconfiguration \nof all elements of the system.