Possibilistic Decision Making in Sensor Systems

This paper considers an alternative to the Bayesian approach to decision making that is based on possibility theory. The possibilistic method uses a minimax criterion to choose the least risky action based on the preference and possibility of the outcome of an action after evaluation of the measured sensor data. The advantage of the possibilistic method when compared with the Bayesian method is that it requires only an ordinal ranking of the cost associated with each action and the uncertainty about the state of the exærnal world. Owing to its qualitative character, the possibilistic decision maker is less sensitive to inaccuracies in a piort knowledge and cost estimates than the Bayesian decision maker at the expense of a degraded performance in situations where accurate a priori knowledge and cost estimetes are available. This robstness of the possibilistic decision maker to inaccuracies in a priori knowledge and cost estimates is demonstrated in a case study where an average cost criterion is used to compare the performance of a possibilistic and a Bayesian decision maker. It is also shown that in a scenario with an intelligent opponent the possibilistic decision maker offers a lower expected cost than a Bayesian decision maker