In GPS-denied environments, robot systems typically revert to navigating with dead-reckoning and relative mapping, accumulating error in their global pose estimate. In this paper, we propose Feature-based Localization between Air and Ground (FLAG), a method for computing global position updates by matching features observed from ground to features in an aerial image. Our method uses stable, descriptorless features associated with vertical structure in the environment around a ground robot in previously unmapped areas, referencing only overhead imagery, without GPS. Multiple-hypothesis data association with a particle filter enables efficient recovery from data association error and odometry uncertainty. We implement a stereo system to demonstrate our vertical feature based global positioning approach in both indoor and outdoor scenarios, and show comparable performance to laser-scan-matching results in both environments.

The ability to refuel satellites on-orbit using teleoperated robots could extend the life of satellite missions, potentially saving time and resources for both industry and space agencies. One critical step in the refueling process is to gain access to the satellite's fueling port by cutting through a seam of tape that adheres two sections of multi-layer insulation (MLI) covering the satellite body. The deformable, delicate, specular nature of the metalized tape used to adhere sections of MLI to one another makes cutting the tape seams a difficult task to perform remotely, a difficulty that is compounded by unavoidable multi-second communications delays. Virtual-fixture-assisted teleoperation can help mitigate some of these issues by constraining the motion of the cutting blade based on a priori knowledge of the task. Adding force control may further assist teleoperators by automatically setting a desired normal cutting force. This paper presents the results of a 20-participant user study for a representative satellite tape-cutting task, showing that force control based on a virtual fixture mapped to the MLI blanket plane can decrease both tape bunching and operator workload.

Direct teleoperation with multisecond time-delayed telemetry between master and slave is challenging for humans to perform. When controlling a holonomic robot with many degrees of freedom, operators may incidentally provide commands in an intended direction without realizing their mistake until receiving feedback several seconds later. For some applications, imposing a virtual nonholonomic constraint (VNHC) on the motion of the end effector can help prevent operators from moving in an unintended direction by reducing the number of controllable degrees of freedom. This paper presents the development of a VNHC for a planar time-delayed telerobotic task, motivated by an on-orbit telerobotic satellite servicing operation. We also describe a nonholonomic virtual fixture (NHVF) that adheres to the VNHC to further reduce the potential for operators to input mistaken commands. We report the results of a pilot study in which teleoperation with a VNHC was found to have comparable task performance to holonomic planar teleoperation, while decreasing operator workload. The NHVF was found to decrease performance slightly, though user feedback indicated that a differently implemented virtual fixture and controller may improve performance.

Humans are an integral part of many tasks performed by mobile robots and human operator ability can vary greatly between users. Understanding operator skill level can allow for user interfaces to adapt giving novice users additional assistance and expert users more freedom. In this paper, a method is proposed for predicting a human teleoperator's performance based on user behavior observed during the first few minutes of teleoperation. Preliminary analysis indicates a trend between operator performance and the proportion of time an operator spends thinking during a particular segment of time.