RESULTS IN SLIDING AUTONOMY FOR MULTI-ROBOT SPATIAL ASSEMBLY

摘要

We have developed a software architecture for teams ofrnrobots and humans to jointly perform tightly coordinatedrntasks, such as assembly of structures in orbit or on planetaryrnsurfaces. While we envision that robots will autonomouslyrnperform such work in the future, the state ofrnthe art falls short of the capabilities necessary to handlernall possible contingencies. Our architecture provides arnprincipled methodology for human involvement to optimizernboth task efficiency and robustness by combiningrnrobot capability with human intuition. We call suchrnmixed control strategies “sliding autonomy”. Robots accomplishrnas many of the tasks as they can autonomously,rnand human operators take over control to perform thosernthat cannot be easily automated or to provide help whenrnthe robots fail. In this paper, we discuss results from recentrnexperiments that quantify the effect of different levelsrnof autonomy on the system's overall performance. Byrnintroducing two modes of sliding autonomy, we are ablernto achieve the high reliability of a teleoperated systemrncombined with the high efficiency of autonomous operation.rnThe incurred mental demand of the operator is directlyrnproportional to the increase in system efficiency.