TASK-DIRECTED COMPUTATION OF QUALITATIVE DECISIONS FROM SENSOR DATA

This article describes a novel approach to sensor-based decision makin g based on formulating and solving large systems of parametric constra ints. The constraints describe both a model for sensor data and the cr iteria for correct decisions about the data. An incremental constraint solving technique that performs decision-directed model recovery is d eveloped. This method is straightforward to apply, is easily paralleli zed, and convergence can be demonstrated under very reasonable structu ral and statistical assumptions. This approach is demonstrated on seve ral different decision-making problems involving manipulation and cate gorization of objects observed with a range scanner. The experiments i ndicate that simultaneous solution of both model constraints and decis ion criteria can lead to efficient and effective decision making, even when the observed data does not strongly determine a data model.