Self-Corrective Autonomous Systems using Optimization Processes for Detection and Correction of Unexpected Error Conditions

Abstract

A theoretical framework for autonomous self-detection and self-correction of unexpected error conditions is derived by incorporating the principles of operation in autonomous control in biological evolution.  Using the biologically inspired principles, the time-dependent multi-dimensional disparity vector is used as a quantitative metric for detecting unexpected and unforeseeable error conditions without any external assistance.  The disparity vector is a measure of the discrepancy between the expected outcome predicted by the autonomous system and the actual outcome in the real world.  It is used as a measure to detect any unexpected or unforeseeable errors.  The process for autonomous self-correction of the self-discovered errors is an optimization process to minimize the errors represented by the disparity vectors.  The strategies for prioritizing the urgency of corrective actions are also provided in the theoretical derivations.  The criteria for any change in direction of the corrective actions are also provided quantitatively.  The criteria for the detection of the minimization and maximization of errors are also provided in the autonomous optimization process.  The biological correspondences of the emotional responses in relation to the autonomic self-corrective feedback systems are also provided.