The sensor information that MI (mode identification) can passively acquire is not always adequate to allow an unambiguous diagnosis, which may make it impossible for MR to identify the appropriate recovery actions. In general, it may be ambiguous which of two failures has occurred if both are consistent with observations, and it may be ambiguous whether a given component or the sensor responsible for measuring that component has failed. If an encoder indicates that a wheel drive motor is not turning, or is turning too slowly, that could be a sensor error: the encoder may be skipping counts or go entirely dead. While MI could look at other evidence, such as motor currents, to see if the motor appears to be stalled or encountering unusually high torque, that evidence may not be sufficient to determine the true fault. If the encoder is giving low readings, the PID (proportional, integral and differential) controller, which tries to maintain the desired encoder values, will tend to speed up the motor, resulting in higher wheel currents, which could be taken as evidence of excessive torque. Furthermore, failures such as a motor stall or excessive torque can reflect a number of different underlying faults, such as seized bearings or a rock caught in the wheel.

The true state of the rover may be determined by performing experiments designed to eliminate certain candidate diagnoses. For example, if the wheel does not appear to be turning, the rover could try backing up to see if the wheel is caught on a rock. Or it could try driving with some subset of the wheels, and use other sensors to determine if the rover is moving in a manner consistent with the encoder readings. To deal with these ambiguities, we are adding the capability to perform active sensing and testing in order to narrow the candidate situation assessments (diagnoses) and in order to evaluate the utility of alternative recovery plans. In support of this active testing, MIR can make use of its models, both to determine when there are multiple competing diagnosis and to identify activities it can perform that will rule out or confirm certain hypotheses. Reasoning about the information to be gained by executing actions exceeds the ability of the MIR system designed for the Remote Agent, but we are working to provide that capability.

Work in active testing for diagnosis is typically based on probe selection for circuit diagnosis, and it relies on certain simplifying assumptions that are valid for circuits but not for rovers. Some of the key assumptions are:

1. Measurements do not affect the state of the system being diagnosed.
2. All measurements have equal cost.
3. The goal of making measurements is to eliminate ambiguity as quickly as possible (i.e., to minimize the total number of measurements); the order of measurements is otherwise irrelevant.

These assumptions lead to a minimum entropy measure for probe selection. The next probe selected is the one that results in the lowest expected entropy of the probability distribution of diagnoses. This policy tends to minimize the total cost of measurements, under the assumptions listed above. However, these assumptions do not hold in the rover domain, for the following reasons, so minimizing entropy is not sufficient.

1. Any information that can be obtained without changing the state of the rover, as long as it is not too expensive to compute, is already continuously available to MI. Any additional tests involve causal action, such as spinning a wheel or taking a picture from a camera.
2. On a rover, some sensing actions may have very high cost, including the possibility of causing some undesirable side effects, while others are relatively cheap.
3. In the rover autonomy architecture, the main purpose of diagnosis is to disambiguate the rover state enough to find an appropriate recovery plan. Thus, not all ambiguities are equal: the value of information depends on the value of the recovery it supports. In the case of multiple faults, one fault may be more critical and need immediate response, meaning measurements relevant to that fault have priority. If several candidate faults have the same recovery procedure, fully disambiguating the fault may even be unimportant. We are exploring a modification of the minimum-entropy model, which ranks measurements according to the recovery actions they support and penalizes measurements based on the cost of the corresponding sensing actions.