Jet Propulsion Laboratory
Pennsylvania State University
Autonomous systems have become critical to the success of military and scientific missions. Vehicles like the Mars Exploration Rovers, which can autonomously drive through a cluttered environment to a goal and autonomously identify and extract features of scientific interest (e.g., dust devils and clouds) from images taken by onboard cameras, and the Boeing X-45A unmanned air vehicle (UAV), which demonstrated the first autonomous flight of a high-performance, combat-capable UAV and the first autonomous multi-vehicle coordinated flight, have reduced the level of human intervention from inner-loop control to high-level supervision.
However, human involvement is still a critical component of robotic systems. In some cases, it is necessary from a legal and arguably moral standpoint (e.g., in autonomous strike missions), but in most cases humans are necessary because of the limitations of current technology. For example, it is still impossible for a robot to navigate autonomously along a crowded sidewalk or for a robotic explorer to demonstrate initiative or “decide what is interesting.” Even recovering from an error, such as a stuck wheel or an actuator fault, generally requires human intervention.
The presentations in this session focus on aspects of autonomy that will bring robotic systems from controlled devices that can function for a few minutes without human intervention to systems that can function autonomously for days or weeks in poorly characterized, or even unknown, environments. The speakers, who represent academia, government, and industry, cover both aeronautical and space autonomous systems. Their presentations have been organized to progress from a single vehicle (including human interaction with the vehicle) to teams of
robots to the incorporation of autonomous unmanned air systems into the National Air Transportation System.
The first talk, by Mark Campbell (Cornell University), focuses on (1) techniques for enabling “intelligence” in autonomous systems through probabilistic models of the environment and (2) the integration of human operators into the control/planning loop. Chad Frost (NASA Ames Research Center) provides an overview of the challenges to increasing automation in NASA’s current and future space missions, highlights examples of successful autonomous systems, and discusses some of the lessons learned from those experiences.
The subject of the third talk, by Stefan Bieniawski (Boeing Research and Technology), is the role of health awareness in multi-vehicle autonomous systems. He describes how such systems can address failures of components in a vehicle or the failure of a vehicle in the team. In the final presentation, Ella Atkins (University of Michigan) discusses the formidable challenges associated with the safe, efficient integration of unmanned air systems into airspace currently traveled by manned aircraft and the importance of automation and autonomy in the deployment of the next-generation air transportation system (NextGen).