NASA has identified a real need to innovate fault management operational techniques. In particular, there is need for techniques that can enable greater autonomy in space missions, and to reduce the need to “safe” a system in case of anomalous environmental or internal behaviour. Of particular importance is innovation that allows for a dramatic increase in the autonomy of robotics required to sustain and support human exploration/habitation of the moon and Mars.
We propose to leverage a variety of causal inference techniques to optimize in-the-loop fault management operations in order to enable autonomous mobile robots operating in hazardous locations such as the moon and Mars. A major focus is to improve the Technical Readiness Level of Autonomous Mobile Robots (AMR’s) that play a significant role in lunar ice water mining in the Permanently Shadowed Regions (PSR’s) of the moon.
Specifically, we propose a framework/methodology to automate the application of the Structural Causal Model (SCM) pioneered by computer scientist Judea Pearl in real-time and near real-time applications to improve AMR autonomy.
There are a wide variety of NASA applications due to the heavy reliance on robotics/control systems to enable NASA missions in space and on other planets. The targetted use case we would like to focus on in the proposal is the advancement of autonomy of Autonomous Mobile Robots (rovers) and their ability to operate more effectively in unexplored domains. Particularly as NASA begins to consider the use of AMR's to do more than mere scientific exploration, but looks at them as a method to enable human exploration and habitation of the moon/Mars.
A framework that enables increased AMR autonomy would be useful in a variety of earthbound industries including but not limited too:
1.) inspections of industrial assets, such as remote solar/wind farms that is currently conducted by humans or not at all.
2.) Enabling more autonomy in agricultural operations.
3.) Disaster Response, due to the ability to navigate uncertain terrain, scenarios.