Establishing a link between multiple-sensor outputs and non-deterministic decision-making

In many surveillance systems, an operator will have to make a decision on whether a target is present based on one or more sensor readings. For the latter, when an operator has at his disposal two or more sensor outputs his decision on a whether a target is present or not is likely to be influenced by those sensor readings. For instance, when both sensors report, the operator is more likely than not to decide a target is present, he is less likely to decide a target is present when only one sensor reports, and is quite probably very unlikely to deem a target to exist when neither sensor makes a report. In addition, studying the likely behaviour and biases of an operator (and positively using them) for a given scenario might take a long time due to operator availability and/or ethics issues. In this case, we can make use of beta distributions to generate random numbers that represent the likely probability of a decision given sensor outputs. Thus, the purpose of this paper is two-fold: 1) We wish to study the relationship between an operator's non-deterministic decision strategy and multiple sensor readings and 2) Given a set of possible operators and their decision biases, we wish to choose the most suitable in terms of mitigating bad choices or `risk'.