Analysis of science team activities during the 1999 Marsokhod Rover Field Experiment: Implications for automated planetary surface exploration

This work analyzes the behavior and effectiveness of a science team using t he Marsokhod mobile robot to explore the Silver Lake region in the Mojave D esert near Baker, California. The work addresses the manner in which the ge ologists organized themselves, how they allocated their time in different a ctivities, how they formed and communicated scientific hypotheses, and the frequency with which they requested different types of data from the missio n archive during the first 3 days of the mission. Eleven scientists from th e NASA Ames Research Center and three of the five scientists who participat ed from their home institutions were videotaped as they worked throughout t he 3-day experiment. The videotape record indicates that 46% of available p erson-hours were consumed in semistructured or formal meetings and that onl y 1% of their time was spent studying immersive, three-dimensional virtual reality models of the robot's surroundings. The remainder of their time was spent in unstructured work sessions in groups Of two or three. Hypothesis formation and evolution patterns show a meager flow of information from the distributed science team to the on-site team and a bias against reporting speculative hypotheses. Analysis of the visual imagery received from the ro bot indicates that acquisition of the large panoramic information leads to high levels of redundancy in the data acquired. The scientists' archive req uests indicate that small, specifically requested image targets were the mo st frequently accessed information. The work suggests alternative organizat ional structures that would expedite the flow of information within the geo logic team. It also advocates emphasizing specific science targets over hig h-resolution, stereoscopic, panoramic imaging when programming a mobile rob ot's onboard cameras.