This article was based on the work carried out by . In future NASA’s space missions will include more and more interactive robots. The Curiosity rover that has been recently sent to Mars was a good example for that. These kinds of robots require new remote operation mechanisms for effective use. In such a tele-operated context, a human team should constantly supervise the robot and manually perform tasks whenever needed.
An important aspect of such operations is the ability to allocate tasks between humans and robots effectively. This capability was known as Adjustable Autonomy (Adaptive Autonomy) so that the automation can be smart enough to achieve the autonomy required according to changing situations. Human-robot interactions are closely related to adjustable autonomy, and they both go side-by-side. Apparently, human robot operations are highly dependent upon the scenario so that they become specific to a given robot, thus making it hard to generalize them. Given below is a sub-set of such human robot operations.
- Physical structure assembly
- Object transportation
- Rescue operations support
- Process control
- Space-craft system control
Adjustable autonomy has been applied for these operations so that shifting the autonomy between humans and robots harnessed better results. That way, human intervention can aid robots to perform better in the presence of uncertainties or problems.
NASA claims that minimizing the human interaction in robotic operations might not translate to better performance. For instance, tele-operating a mobile robot in an unknown or unstructured planetary surface will be more time efficient than an autonomous operation. This article describes some of the performance metrics that can be used to evaluate the team performance and success rate in human-robot interactions.
The study described in  demonstrated their work based on Robotic Reconnaissance Operations. Robotic reconnaissance operations involve remotely controlling a rover to scout on an unknown planetary terrain prior to astronaut extra vehicular activity. This particular operation was carried out with adjustable autonomy and the performance was measured based on Task Success, Team Performance, and Team workload.
Most importantly this work highlighted that the task success rate was higher when used human intervention on difficult terrains. This is because the robot could not decide which path to travel in such conditions.
Therefore studies like these allow us to conclude that Human-Robot systems would surpass the performance of fully-autonomous schemes, in unknown or unpredictable conditions, and the future research in this arena will focus on that.
 Debra Schreckenghost, Tod Milam, and Terrence W. Fong “Measuring Performance in Real Time during Remote Human-Robot Operations with Adjustable Autonomy” IEEE Intelligent Systems, Vol. 25, No. 5, September, 2010, pp. 36-44.