Integrated Topological Planning and Scheduling
29
approach, namely, we assume a set of pre-determined workstation locations and
capabilities, and we aim to select a subset thereof that optimizes some objective
function, such as e.g. minimization of travel time. We believe that this approach
is more advantageous for application to real-world factory floors, where there
are strict constraints that are often tacit in the existing factory floor layout.
In the proposed problem definition, a set of workstations is topologically
located within a factory floor. One or more agents (humans, robots) may be
assigned to work at one or more workstations. Each agent has unique skills that
can be used to carry out tasks. Some of those tasks are localized to workstations,
while others require transport of materials or workpieces between workstations
and other target locations within the factory. We assume that the factory processes workpieces according to known processes that involve a series of tasks to
be performed in sequence. Each workstation is differentiated by assigned agents
and assigned workpiece types (and thus process types). A single workpiece is
processed in one workstation and no workstation changes are allowed. However,
distinct workpieces of the same type may be selected to be processed in more
than one workstations, always obeying processing of single workpiece to single
station.
For the first step of the process, we consider the following entities:
– Set of workstations W , located within a factory floor with pose defined as
{x wi , y wi , θ wi }∀w i ∈ W , and each with a set of points of interest (PoI) corresponding to relevant locations in the workstation, such as worker location,
robot manipulator base point and AGV dropoff locations.
– Set of processes P that may be carried out by completing specified steps
(tasks) for each process.
– Set of tasks T , each of which corresponds to an elementary action that can
be performed by an agent and advances the process at hand. t i ∈ p j , ∀t i ∈
T, p j ∈ P
– Set of agents A, each of which corresponds to a worker or robot and is defined
by a series of skills.
– Set of skills S indicating the skill of an agent a at task t, where ∀t i ∈ T and
a j ∈ A ∃s ij ∈ S
– Set of target locations within the factory L with location {x li , y li }∀l i ∈ L
– A set of desired throughput rates per process Q P , which denote the complete
process iterations per process type, per hour, that the factory should achieve.
Given the above entities definition, a solution to the topology and task assignment problem comprises the following elements:
– A matrix S aw of agent-workstation assignments.
– A matrix R pw of process-workstation assignments.
It is noted that even though potential agent to task assignments are computed
during the evaluation process, they are not part of the solution, as precise task
assignments will be performed as part of the second stage, that is, the detailed
process scheduling.
29
approach, namely, we assume a set of pre-determined workstation locations and
capabilities, and we aim to select a subset thereof that optimizes some objective
function, such as e.g. minimization of travel time. We believe that this approach
is more advantageous for application to real-world factory floors, where there
are strict constraints that are often tacit in the existing factory floor layout.
In the proposed problem definition, a set of workstations is topologically
located within a factory floor. One or more agents (humans, robots) may be
assigned to work at one or more workstations. Each agent has unique skills that
can be used to carry out tasks. Some of those tasks are localized to workstations,
while others require transport of materials or workpieces between workstations
and other target locations within the factory. We assume that the factory processes workpieces according to known processes that involve a series of tasks to
be performed in sequence. Each workstation is differentiated by assigned agents
and assigned workpiece types (and thus process types). A single workpiece is
processed in one workstation and no workstation changes are allowed. However,
distinct workpieces of the same type may be selected to be processed in more
than one workstations, always obeying processing of single workpiece to single
station.
For the first step of the process, we consider the following entities:
– Set of workstations W , located within a factory floor with pose defined as
{x wi , y wi , θ wi }∀w i ∈ W , and each with a set of points of interest (PoI) corresponding to relevant locations in the workstation, such as worker location,
robot manipulator base point and AGV dropoff locations.
– Set of processes P that may be carried out by completing specified steps
(tasks) for each process.
– Set of tasks T , each of which corresponds to an elementary action that can
be performed by an agent and advances the process at hand. t i ∈ p j , ∀t i ∈
T, p j ∈ P
– Set of agents A, each of which corresponds to a worker or robot and is defined
by a series of skills.
– Set of skills S indicating the skill of an agent a at task t, where ∀t i ∈ T and
a j ∈ A ∃s ij ∈ S
– Set of target locations within the factory L with location {x li , y li }∀l i ∈ L
– A set of desired throughput rates per process Q P , which denote the complete
process iterations per process type, per hour, that the factory should achieve.
Given the above entities definition, a solution to the topology and task assignment problem comprises the following elements:
– A matrix S aw of agent-workstation assignments.
– A matrix R pw of process-workstation assignments.
It is noted that even though potential agent to task assignments are computed
during the evaluation process, they are not part of the solution, as precise task
assignments will be performed as part of the second stage, that is, the detailed
process scheduling.