The crewed exploration of Moon and Mars requires the construction and maintenance of infrastructure on the alien surfaces before a crew arrives. Robotic coworkers are envisioned to take over the physical labor required to set-up crew habitats, energy supplies, and return vehicles in the hazardous environment. Deploying these robots in such a remote location poses a challenge that requires autonomous robot capabilities in combination with effective Human Robot Interfaces (HRIs), which comply with the harsh conditions of deep space operations. An astronaut-robot teleoperation concept targeting these topics has been evaluated in DLR and ESA's METERON SUPVIS Justin experiment where astronauts on-board the International Space Station (ISS) commanded DLR's humanoid robot Rollin' Justin in a simulated Martian environment on Earth. This work extends on our previously presented approach to supervised autonomy. It examines the results of the two follow-up experiment sessions which investigated maintenance and assembly tasks in real-world scenarios. We discuss the use of our system in real space-to-ground deployment and analyze key performance metrics of the HRI and the feedback given by the astronauts.

Haptic guidance is a promising method for assisting an operator in solving robotic remote operation tasks. It can be implemented through different methods, such as virtual fixtures, where a predefined trajectory is used to generate guidance forces, or interactive guidance, where sensor measurements are used to assist the operator in real-time. During the last years, the use of learning from demonstration (LfD) has been proposed to perform interactive guidance based on simple tasks that are usually composed of a single stage. However, it would be desirable to improve this approach to solve complex tasks composed of several stages or gestures. This paper extends the LfD approach for object telemanipulation where the task to be solved is divided into a set of gestures that need to be detected. Thus, each gesture is previously trained and encoded within a Gaussian mixture model using LfD, and stored in a gesture library. During telemanipulation, depending on the sensory information, the gesture that is being carried out is recognized using the same LfD trained model for haptic guidance. The method was experimentally verified in a teleoperated peg-in-hole insertion task. A KUKA LWR4+ lightweight robot was remotely controlled with a Sigma.7 haptic device with LfD-based shared control. Finally, a comparison was carried out to evaluate the performance of Gaussian mixture models with a well-established gesture recognition method, continuous hidden Markov models, for the same task. Results show that the Gaussian mixture models (GMM)-based method slightly improves the success rate, with lower training and recognition processing times.

As the human race expands its horizon toward a multiplanetary existence, infrastructures on the target planets have to be constructed and maintained to pave the way for humans. The support of robotic coworkers plays a key role in setting up habitats, energy supplies, and return vehicles, until the completion of such infrastructures in the hazardous planetary environment. The operation of these robots require capabilities including autonomy, communication, and human-robot interface design to meet the challenges of the harsh conditions in space deployment. This letter examines these topics through German Aerospace Center (DLR) and European Space Agency's METERON SUPVIS Justin space telerobotics experiments, during that astronauts on-board the International Space Station command DLR's humanoid robot Rollin' Justin to survey and maintain a simulated Martian solar farm on Earth. Based on the first experiments conducted on August 25, 2017, this letter discusses several astronaut-robot collaboration concepts in real space-to-ground deployment and provides preliminary insights for future manned Mars missions.

In this paper, we propose a closed-loop force sensor based nested admittance/impedance control strategy to actively estimate and minimize the effects of geometric misalignment that naturally occur during assembly tasks with compliant robots. The method allows the robot to be used with a stiff impedance control setting, which is beneficial for free air motion performance, yet allows to adjust for large misalignment errors between parts that need be assembled.

This paper introduces a new Learning from Demonstration (LfD)-based method that makes usage of robot effector forces and torques recorded during expert demonstrations, to generate force-based haptic guidance reference trajectories on-line, that are intended to be used during haptic shared control for additional operator 'guidance'. Derived haptic guidance trajectories are superimposed to master-device inputs and feedback forces within a bilateral control experiment, to assist an operator by the guidance during peg-in-hole insertion. We show that 96 peg-in-hole expert demonstrations were sufficient to obtain a good model of the task, which was used on-line to generate haptic guidance trajectories in real-time with a 1kHz sampling rate.

On July 28th 2014, 23:47 UTC, the European Space Agency launched the Haptics-1 Kit to the International Space Station (ISS) on its last Automated Transfer Vehicle ATV-5. The Kit reached the station two weeks later, marking the first haptic master device to enter the ISS. The first force-feedback and human perceptual motor performance tests started to take place on December 30th 2014, and are the first of their kind in the history of spaceflight. Three astronauts participated in the Haptics-1 experiment until November 2015, allowing the investigation of the effects of microgravity on various psycho-motor performance metrics related with the usage of haptic feedback. Experiments are conducted following full adaptation to the space environment (after 3 months in space). This paper introduces the Haptics-1 experiment and associated hardware. Detailed experimental results are reported from a first stiffness just noticeable difference (JND) experimental study in space, carried out on the ISS and pre-flight on ground with 3 astronauts. The first findings from the experiment show no major alterations in-flight, when compared to on-ground data, if the manipulandum is secured in flight against a sufficiently stiff reference structure.

This paper introduces a new Learning from Demonstration (LfD)-based method that makes usage of robot effector forces and torques recorded during expert demonstrations, to generate force-based haptic guidance reference trajectories on-line, that are intended to be used during haptic shared control for additional operator 'guidance'. Derived haptic guidance trajectories are superimposed to master-device inputs and feedback forces within a bilateral control experiment, to assist an operator by the guidance during peg-in-hole insertion. We show that 96 peg-in-hole expert demonstrations were sufficient to obtain a good model of the task, which was used on-line to generate haptic guidance trajectories in real-time with a 1kHz sampling rate.

In a typical space teleoperation task, mismatches between the viewing direction of the operator and the direction of their required control input are often unavoidable. To execute these tasks, the operator is then required to perform mental rotations. Recent studies have shown that the task performance can thereby significantly decrease. In this paper, for the first time, the influence of mental rotations on task performance is studied if hap tic feedback is provided to the operator. A human factors experiment is conducted which analyses the influence of two different hap tic feedback control methods via various visual missmatch angles. The rotation is thereby set to the extreme cases of 0. and 180.To clearly analyze the effects. The first hap tic feedback method consists of direct, scaled force and torque feedback to the operator as measured by a force/torque sensor at the slave robot. The second method consists of hap tic shared control which provides artificially generated guidance forces to the operator. It is shown that mental rotations decrease teleoperation performance despite the addition of direct force feedback. In contrast, hap tic shared control provides lower increase in the operator mental workload and also less between-operator variability of errors made due to the mental rotations.

The CyberGrasp™ is a well known dataglove-exoskeleton device combination that allows to render haptic feedback to the human fingers. Its design, however, restricts its usability for teleoperation through a limited control bandwidth and position sensor resolution. Therefore the system is restricted to low achievable contact stiffness and feedback gain magnitudes in haptic rendering. Moreover, the system prohibits simple adaption of its controller implementation.