24
I. Chatzikonstantinou et al.
that cannot be adequately tackled through automation of teams of robots alone.
Shortcomings in perception, reasoning or actuation of robots, compared to their
human counterparts, limit their applicability to tasks requiring high level of
corresponding abilities. Combination of robots with humans results in versatile
teams and allows achieving the best of both worlds: HRC allows robots to undertake tedious and dangerous tasks from humans [1,2]. At the same time humans
assist, supervise and take over tasks that are challenging for robots, or in case
of failure.
As an illustrative scenario, the automation of recycling factory floors processing Waste Electrical and Electronic Equipment (WEEE) is mentioned. In
such a scenario, a stream of highly heterogeneous end-of-life devices need to be
skillfully manipulated in order to extract and sort critical components that are
valuable or possess dangerous substances. For a robot, this task presents a significant challenge as it requires increased perception and actuation capabilities.
An HRC setup in this case may enable the use of robots in a scenario that would
otherwise be exclusive.
In an HRC scenario, one aspect that has vital importance is that of task
planning and scheduling. The coordination of single-robot, single-human teams,
or HRC teams comprising few robot and human agents has been extensively
studied in literature [3–6]. In this scenario, collaboration is on a single main
task, e.g. the assembly or disassembly of a single device, and the goal is to
assign to agents and sequence operations so as to reach the goal of the task (i.e.
satisfy all constraints) while also minimizing a problem-defined objective. There
is usually a single workspace that is shared amongst agents, where all tasks are
processed.
On the contrary, processes taking place in modern industrial shop floor are on
a different scale. The amount of active agents including (including humans and
robots) is much larger, and is organized into several smaller teams that process
tasks in parallel. Another distinguishing feature is that processes take place in
several workstations, which are topologically organized and may be unique in
terms of capabilities and tooling. Topology, in fact, is an aspect that is of great
importance to achieving a comprehensive orchestration in the large scale, and
one that can directly affect efficiency [7,16].
It is evident through the above that the orchestration of HRC teams at
the factory scale is a unique problem. Orchestration solutions for large-scale
HRC orchestration need to account for the problem’s unique properties, such as
limitations in robot manipulation abilities, sharing of human resources within
the factory floor, non-holonomic constraints in mobile robots etc. This problem
is one that has received less attention so far in literature. Nonetheless, efficient
orchestration of large scale human-robot collaborative teams is key to achieving
factory-wide efficiency [7].
1.1 Contribution
This paper considers the problem of efficiently orchestrating large HRC teams,
in industrial settings, to perform tasks that are characterized by collaboration
I. Chatzikonstantinou et al.
that cannot be adequately tackled through automation of teams of robots alone.
Shortcomings in perception, reasoning or actuation of robots, compared to their
human counterparts, limit their applicability to tasks requiring high level of
corresponding abilities. Combination of robots with humans results in versatile
teams and allows achieving the best of both worlds: HRC allows robots to undertake tedious and dangerous tasks from humans [1,2]. At the same time humans
assist, supervise and take over tasks that are challenging for robots, or in case
of failure.
As an illustrative scenario, the automation of recycling factory floors processing Waste Electrical and Electronic Equipment (WEEE) is mentioned. In
such a scenario, a stream of highly heterogeneous end-of-life devices need to be
skillfully manipulated in order to extract and sort critical components that are
valuable or possess dangerous substances. For a robot, this task presents a significant challenge as it requires increased perception and actuation capabilities.
An HRC setup in this case may enable the use of robots in a scenario that would
otherwise be exclusive.
In an HRC scenario, one aspect that has vital importance is that of task
planning and scheduling. The coordination of single-robot, single-human teams,
or HRC teams comprising few robot and human agents has been extensively
studied in literature [3–6]. In this scenario, collaboration is on a single main
task, e.g. the assembly or disassembly of a single device, and the goal is to
assign to agents and sequence operations so as to reach the goal of the task (i.e.
satisfy all constraints) while also minimizing a problem-defined objective. There
is usually a single workspace that is shared amongst agents, where all tasks are
processed.
On the contrary, processes taking place in modern industrial shop floor are on
a different scale. The amount of active agents including (including humans and
robots) is much larger, and is organized into several smaller teams that process
tasks in parallel. Another distinguishing feature is that processes take place in
several workstations, which are topologically organized and may be unique in
terms of capabilities and tooling. Topology, in fact, is an aspect that is of great
importance to achieving a comprehensive orchestration in the large scale, and
one that can directly affect efficiency [7,16].
It is evident through the above that the orchestration of HRC teams at
the factory scale is a unique problem. Orchestration solutions for large-scale
HRC orchestration need to account for the problem’s unique properties, such as
limitations in robot manipulation abilities, sharing of human resources within
the factory floor, non-holonomic constraints in mobile robots etc. This problem
is one that has received less attention so far in literature. Nonetheless, efficient
orchestration of large scale human-robot collaborative teams is key to achieving
factory-wide efficiency [7].
1.1 Contribution
This paper considers the problem of efficiently orchestrating large HRC teams,
in industrial settings, to perform tasks that are characterized by collaboration