Transparent interaction and human-robot collaboration for military operations

ROBOTS ARE IMPORTANT IN THE MILITARY?

THEY DO WHAT SOLDIERS CANNOT

THEY INCREASE THE SOLDIERS’ SCOPE OF INFLUENCE

clearing buildings;
search and rescue in disaster areas and battlefields;
detonation and disposal of explosives;
reconnaissance and surveillance.

gathering data to support soldiers’ situation awareness,
transporting soldier equipment,
distributing supplies to soldiers in the most forward resupply positions,
facilitating commanders’ decision-making (collecting, organizing, and prioritizing data)
keeping soldier’s safe by providing greater stand-off distance from the enemy for maneuvers and convoys

soldier completes all the tasks

knowledge, and all the other actions needed for interdependent operation

TEAMWORK

Robots are already being designed to do taskwork

Team behaviors, such as communication and coordination, can be simulated by robots to support important facets of HRI, such as mutual predictability and shared knowledge

Transparent HRI

Transparency is an emergent property of the HRI process whereby a human operator has a clear and accurate understanding of how the robot gathers information, processes that info, and makes decisions

the world, itself, and its situation to independently compose and select among different courses of action to accomplish its goals

adaptive automation
(the level of automation is changeable by the system)
+
adjustable automation
(the level of automation is changeable by an external operator or system)

a human (i.e., teleoperator) can mechanically manipulate items or sense objects at a different location than where they are currently located, using a mechanical or robotic apparatus

an operator manages multiple robots by interacting with them individually

TELEOPERATION

SUPERVISORY CONTROL

MIXED-INITIATIVE SYSTEMS

Challenges: operators experience issues related to cognitive tunneling, decreased field of vision, degraded sense of spatial orientation, attention switching, and motion sickness

Challenges: as the number of robots being supervised increases, the operators’ workload increases, their situation awareness decreases, their response times increase, the number of tasks that can be successfully completed within a designated time interval decreases, and the number of system failures and accidents increase

Challenges: mode confusion – when an operator believes the automation is in a different mode than it currently is

This combination of independence and flexibility is particularly desirable in complex, dynamic environments, the kind in which the US military anticipates operating

depends on a mutual understanding of what all the team members are capable of doing, what they are responsible for doing, and what they are currently doing

TEAMWORK AUTONOMOUS ROBOTS+HUMAN

Human does better

Machine does better

Both the human and the robot are capable of doing a task suitably

One is better than another in dependence on time

Neither human nor robot can do the task by themselves but can do so jointly as a team

THE BEST VARIANT
A human has to work with a robot that:
supports mixed-initiative interaction,
keeping the human apprised of the overall situation,
soliciting additional information as needed,
completing predefined tasks without human intervention,
dynamically negotiating for initiative when appropriate

TASKS environment

BE IN IT

so long as he is aware of what the automation is doing, how the task is progressing, and can reasonably estimate what the next step or outcome will be

A HUMAN “IN THE LOOP”

“The loop of control” – the interactions between a human operator and automation while conducting one or more tasks

when the human loses awareness of the shared task

A HUMAN “OUT OF THE LOOP”

REASONS

Bad HR communication;
a robot is modeled to be “strong and silent”;
human factor

RESULTS

reduced performance;
reduced situation awareness;
increased automation bias;
undesired actions

can make it difficult for people working with these robots to stay informed about the robot’s actions and the decision making processes that lead to those actions. Thus, we need extensive interaction models for complete understanding between “teammates”

AUDITORY CUES

Synthetic and human voices, tones, alarms

HAPTIC FEEDBACKS

Vibrations and controller torque

SENSOR-BASED DECISION MAKING

A robot sees an environment through sensors and applies some logics on it:
NLP, gestures detection...

INPUT DEVICES

Keyboard, mouse, touch screen

SHARED MENTAL MODELS

Mutual predictability using neuroengineering technologies

In future soldiers and robots will work in mixed teams with shared capabilities

ordinary nurse (COVID-19 hospitals, battlefields, for example);
Mixed-initiative interaction is a promising approach that allows providing qualified assistance to a larger number of hospitalized people with a limited number of medical staff, but considering doctor’s solutions regarding special situation;
We’ll implement the protocol of mixed-initiative interaction in the project.

IMPORTANCE OF THE PAPER FOR THE GROUP’S PROJECT

RoboNurse