HR: 0800h
AN: IN41A-0310    [Abstracts]
TI: A Three Level Autonomous Software System for Increased Science Return
AU: * Robinson, P I
EM: probinson@mail.arc.nasa.gov
AF: SAIC, M/S 269-2 NASA Ames Research Center, Moffett Field, CA 94040 United States
AU: Mancinelli, R L
EM: rmancinelli@mail.arc.nasa.gov
AF: SETI Institute, 515 N. Whisman Road, Mountain View, CA 94043 United States
AU: Landheim, R
EM: rlandheim@mail.arc.nasa.gov
AF: SETI Institute, 515 N. Whisman Road, Mountain View, CA 94043 United States
AB: The development of smart science instruments for autonomous operation (on Earth or in space) has the potential to increase science return and reduce the risk of experiment failure. When researchers are confronted with unexpected data/results of the experimental test system, they must determine whether the experimental setup has failed, or scientific discovery is being made. These two classes of events could have the same time series signature. To directly address this issue, we have developed a three-level software system referred to as E3, which consists of an engineering level, an experiment level, and an executive level. Each level of the software system is designed in a modular fashion using model based feedback controllers. The same feedback control mechanism is used for each level; the model itself determines the level. To determine if failure of the experimental setup can explain the data/results, researchers run calibration tests for hardware (e.g., sensors and actuators) as well as verify that the software (e.g., controls and analog to digital conversion routines) is running as planned. If anomalies are found, then modifications are made to the experimental setup, or the anomaly is accepted as the new baseline state of the instrument. The engineering level of the E3 software system is responsible for this process. To determine if scientific discovery, as opposed to failure, can explain the data/results, a researcher tries to explain the difference between the observed and expected results. These explanations are terms of the basic processes of nature to determine the rate limiting step(s) of a complex set of processes, where the flux could be due to heat transfer, mass transfer, momentum transfer, or chemical reaction processes. Once the differences are understood, modifications are made to the software control of the experiment, as well as to the model the researcher is building over the course of repeated experiments. The experiment level of the E3 software system is responsible for this process. Regardless of whether the experimental setup, or the experiment protocol must be modified in order to maximize science return, it is desired to continue to gather scientific value for an experiment even when it is not going according to plan. In order to accomplish this task, the researcher may fix or float different parameters in the experiment, or modify the experimental conditions to continue to acquire data. The executive level of the E3 software system is responsible for this process. The software system has been developed using a bioreactor - useful for modeling planetary atmospheres, as well as life-support systems. As an example, for the bioreactor studies, the engineering level regulates the performance of the experimental setup, including the bioreactor vessel, temperature control hardware (cooler, heater, lights), light controller (lights), ion-specific electrode hardware, as well as the sample loop hardware. The experiment level regulates the fixed parameters as defined by the researcher (e.g., pH and CO₂, O₂, etc.). The executive level regulates performance of both the engineering and experiment levels to determine which parameters to fix or float in order to maximize the science return.
DE: 0510 Agent-based models
DE: 0594 Instruments and techniques
SC: Earth and Space Science Informatics [IN]
MN: Fall Meeting 2005