Non Deterministic Planning with Evidence and Paradoxical Reasoning Theories

Planetary exploration rovers require high level of autonomy: they should act as much as possible without human intervention. Nevertheless, there are intrinsic uncertainties on activity duration, position of the rover, and other environment characteristics, like soil condition, dust on solar panels, temperature, etc.: disregarding them during planning would bring unreliable plans. A novel, non-deterministic planning approach for autonomous rovers will be presented. Epistemic uncertainties in the models and errors are taken into account in the planning process in order to prevent failures. For each plan, reliability is computed and used to predict the safest one, by means of the Dempster-Shafer Theory of Evidence. In addition, the rover has been endowed with the capability of reallocating its goals. By data-fusing payload and navigation information, it assigns interest values to the existing goals or generates new goals. In this way the planner can choose the most interesting scientific objectives to be analyzed, with limited human intervention, and reallocates its goals autonomously. Dezert-Smarandache Theory of Plausible and Paradoxical Reasoning has been used for information fusion. Finally, some applications to the generation of reliable plans are shown. These tests demonstrate how the planner is able to generate plans that maximize both the reliability and the level of interest