def print_solution(data: VehicleRoutingProblemInstance,
manager: RoutingIndexManager, routing: RoutingModel,
solution: Assignment):
"""Prints assignment on console."""
time_dimension = routing.GetDimensionOrDie('Duration')
total_time = 0
for vehicle_id in range(data.num_plans_to_create):
start_time = None
index = routing.Start(vehicle_id)
plan_output = 'Route for vehicle {}:\n'.format(vehicle_id)
while not routing.IsEnd(index):
time_var = time_dimension.CumulVar(index)
if start_time is None:
start_time = solution.Min(time_var)
plan_output += '{0} Time({1},{2}) -> '.format(
manager.IndexToNode(index), solution.Min(time_var),
solution.Max(time_var))
index = solution.Value(routing.NextVar(index))
time_var = time_dimension.CumulVar(index)
plan_output += '{0} Time({1},{2})\n'.format(
manager.IndexToNode(index), solution.Min(time_var),
solution.Max(time_var))
plan_output += f'Start of the route: {start_time}\n'
plan_output += 'Time of the route: {} sec\n'.format(
solution.Min(time_var) - start_time)
print(plan_output)
total_time += solution.Min(time_var) - start_time
print('Total time of all routes: {} sec'.format(total_time))
def _set_capacity_dimension_constraints(
self, data: VehicleRoutingProblemInstance, routing: RoutingModel,
manager: RoutingIndexManager):
if data.courier_capacities is None:
return
def demand_callback(index):
node = manager.IndexToNode(index)
return data.node_demands[node]
capacity_callback = routing.RegisterUnaryTransitCallback(
demand_callback)
routing.AddDimensionWithVehicleCapacity(
capacity_callback,
0, # waiting time
data.courier_capacities, # maximum distance per vehicle
False, # Force start cumul to zero.
self.CAPACITY_DIMENSION_NAME)
capacity_dimension = routing.GetDimensionOrDie(
self.CAPACITY_DIMENSION_NAME)
for courier_idx, start_utilization in enumerate(
data.start_utilizations):
node_idx = routing.Start(courier_idx)
self._set_constraint_on_var(
node_idx,
TimeWindowConstraint(node=node_idx,
is_hard=True,
from_time=start_utilization,
to_time=start_utilization),
capacity_dimension)
def from_cvrp_solution(cls, data_model: DataModel,
manager: pywrapcp.RoutingIndexManager,
routing: pywrapcp.RoutingModel,
assignment) -> 'Optional[List[VehicleRoute]]':
if not assignment:
return None
res = [] # type: List[VehicleRoute]
for vehicle_id in range(data_model.vehicle_number):
index = routing.Start(vehicle_id)
route_indices = [] # type: List[int]
route_distance = 0 # float
route_load = 0 # float
while not routing.IsEnd(index):
node_index = manager.IndexToNode(index)
route_indices.append(node_index)
route_load += data_model.demands[node_index]
previous_index = index
index = assignment.Value(routing.NextVar(index))
route_distance += routing.GetArcCostForVehicle(
previous_index, index, vehicle_id)
route_indices.append(0) # returns to depot
res.append(
cls(data_model, vehicle_id, route_indices, route_distance,
route_load))
current_app.logger.debug('CVRP Solution:\n{res}')
return res
def draw_route_graph(
data: dict,
routing: RoutingModel,
manager: RoutingIndexManager,
assignment: Assignment,
filename: str = 'route.png',
prog='sfdp',
) -> None:
"""
Draw a route graph based on the solution of the problem.
"""
weights = data['weights']
demands = data['demands']
time_windows = data['time_windows']
graph = pgv.AGraph(directed=True)
def _node(index: int) -> str:
if index == 0:
return f'{index}\nDepot'
return f'{index}\nDemand: {demands[index]}\nRange: {time_windows[index]}'
for vehicle_id in range(data['num_vehicles']):
index = routing.Start(vehicle_id)
while not routing.IsEnd(index):
node_index = manager.IndexToNode(index)
next_index = assignment.Value(routing.NextVar(index))
next_node_index = manager.IndexToNode(next_index)
weight = weights[node_index][next_node_index]
graph.add_edge(_node(node_index), _node(next_node_index), weight=weight, label=weight)
index = next_index
graph.draw(filename, prog=prog)
print(f'The route graph has been saved to {filename}.')
def print_solution(
data: dict,
routing: RoutingModel,
manager: RoutingIndexManager,
assignment: Assignment,
) -> None:
"""
Print the solution.
"""
capacity_dimension = routing.GetDimensionOrDie('Capacity')
time_dimension = routing.GetDimensionOrDie('Time')
total_time = 0
for vehicle_id in range(data['num_vehicles']):
index = routing.Start(vehicle_id)
node_props = []
while not routing.IsEnd(index):
props = node_properties(manager, assignment, capacity_dimension, time_dimension, index)
node_props.append(props)
index = assignment.Value(routing.NextVar(index))
props = node_properties(manager, assignment, capacity_dimension, time_dimension, index)
node_props.append(props)
route_time = assignment.Value(time_dimension.CumulVar(index))
route = "\n -> ".join(['[Node %2s: Load(%s) Time(%2s, %s)]' % prop for prop in node_props])
plan_output = (f'Route for vehicle {vehicle_id}:\n {route}\n'
f'Load of the route: {props[1]}\nTime of the route: {route_time} min\n')
print(plan_output)
total_time += route_time
print(f'Total time of all routes: {total_time} min')
def _set_arc_distance(self, data: VehicleRoutingProblemInstance,
manager: RoutingIndexManager, routing: RoutingModel):
# ========= DISTANCE CONSTRAIN =========
cost_callback_indices = []
# model "cost between nodes per vehicle"
for vehicle_idx in range(data.num_plans_to_create):
def vehicle_cost_callback(from_index, to_index):
from_node = manager.IndexToNode(from_index)
to_node = manager.IndexToNode(to_index)
distance = data.car_distance_matrix[from_node][
to_node] # in meters
return distance
cost_callback_index = routing.RegisterTransitCallback(
vehicle_cost_callback)
cost_callback_indices.append(cost_callback_index)
routing.SetArcCostEvaluatorOfVehicle(cost_callback_index,
vehicle_idx)
routing.AddDimensionWithVehicleTransits(
cost_callback_indices,
0, # waiting time
MAX_TIMESTAMP_VALUE, # maximum distance per vehicle
True, # Force start cumul to zero.
self.DISTANCE_DIMENSION_NAME)
def solve(self, indices_to_visit: List[int] = None) -> Dict[str, Any]:
"""Finds the optimal order of facilities to minimize distance.
Parameters
----------
indices_to_visit : List[int]
The list of indices corresponding to the desired facilities to visit
Returns
-------
Dict[str, Any]
Soltution dictionary with keys:
- 'objective', set to objective value (minified distance)
- 'order', instructions for the order of facilities to visit
"""
if indices_to_visit is None:
indices_to_visit = list(range(len(self.matrix)))
# make sure home location is in the listed, and that the list is sorted
if self.home_index not in indices_to_visit:
indices_to_visit.append(self.home_index)
indices_to_visit.sort()
data = self._create_data_model(indices_to_visit)
# create routing index manager
manager = RoutingIndexManager(len(data['distance_matrix']),
data['num_vehicles'], data['home'])
# create routing model
routing = RoutingModel(manager)
def distance_callback(from_index, to_index):
# returns distance between two nodes
from_node = manager.IndexToNode(from_index)
to_node = manager.IndexToNode(to_index)
dist = data['distance_matrix'][from_node][to_node]
return dist
transit_callback_index = routing.RegisterTransitCallback(
distance_callback)
# define cost of each arc
routing.SetArcCostEvaluatorOfAllVehicles(transit_callback_index)
# set first solution heuristic
search_params = pywrapcp.DefaultRoutingSearchParameters()
search_params.first_solution_strategy = (
routing_enums_pb2.FirstSolutionStrategy.PATH_CHEAPEST_ARC)
# solve problem
assignment = routing.SolveWithParameters(search_params)
return self._extract_solution(manager, routing, assignment,
indices_to_visit)
def print_solution(
data: dict,
routing: RoutingModel,
manager: RoutingIndexManager,
assignment: Assignment,
) -> None:
capacity_dimension = routing.GetDimensionOrDie('Capacity')
time_dimension = None
total_cost = 0
if 'time_windows' in data.keys():
time_dimension = routing.GetDimensionOrDie('Time')
total_time = 0
for vehicle_id in range(data['num_vehicles']):
index = routing.Start(vehicle_id)
node_props = []
while not routing.IsEnd(index):
props = node_properties(data, manager, assignment, capacity_dimension, time_dimension, index)
node_props.append(props)
index = assignment.Value(routing.NextVar(index))
props = node_properties(data, manager, assignment, capacity_dimension, time_dimension, index)
node_props.append(props)
# arc weight
cost = 0
node_indexes = [prop[0] for prop in node_props]
for i in range(len(node_indexes)-1):
idx_from = node_indexes[i]
idx_to = node_indexes[i+1]
cost += data['dist_mat'][idx_from][idx_to]
total_cost += cost
if 'time_windows' in data.keys():
route_time = assignment.Value(time_dimension.CumulVar(index))
route = "\n -> ".join(['[Node %2s(%s)TW(%2s, %2s): Vehicle Load(%2s) and Arrived(%2s)]' % prop for prop in node_props])
plan_output = (f'Route for vehicle {vehicle_id}:\n {route}\n'
f'Cost: {cost}\nLoad: {props[-1]}\nTime: {route_time} min\n')
print(plan_output)
total_time += route_time
else:
route = "\n -> ".join(['[Node %2s(%s): Vehicle Load(%2s)]' % prop for prop in node_props])
plan_output = (f'Route for vehicle {vehicle_id}:\n {route}\n'
f'Cost: {cost}\nLoad: {props[-1]}\n')
print(plan_output)
print(f'Total cost of all routes: {total_cost}\n')
return total_cost
if 'time_windows' in data.keys():
print('*** format ***: \n[Node node_index(demand)TW(Time Window min, max): Vehicle Load(accumulated load) and Arrived(time)]')
print(f'Total time of all routes: {total_time} min')
def _solve(self, routing: RoutingModel,
search_parameters: RoutingSearchParameters,
initial_routes) -> Assignment:
if initial_routes is None:
solution = routing.SolveWithParameters(search_parameters)
else:
routing.CloseModelWithParameters(search_parameters)
initial_solution = routing.ReadAssignmentFromRoutes(
initial_routes, True)
solution = routing.SolveFromAssignmentWithParameters(
initial_solution, search_parameters)
return solution
def add_capacity_constraints(routing: pywrapcp.RoutingModel, data: dict,
demand_callback):
"""
Add capacity constraints.
"""
demand_callback_index = routing.RegisterUnaryTransitCallback(demand_callback)
routing.AddDimensionWithVehicleCapacity(
demand_callback_index,
slack_max=0, # null slack
# vehicle maximum capacities
vehicle_capacities=data['vehicle_capacities'],
fix_start_cumul_to_zero=True, # start cumul to zero
name='Capacity',
)
def apply(self, solver: RoutingModel):
"""Vehicle's occupation is restricted based on demand"""
callback_index = solver.RegisterUnaryTransitCallback(self._callback)
solver.AddDimensionWithVehicleCapacity(
evaluator_index=callback_index,
slack_max=0,
vehicle_capacities=[
vehicle.capacity
for vehicle in self.problem.vehicles.values()
],
fix_start_cumul_to_zero=True,
name='capacity_constraint'
)
def apply(self, manager: pywrapcp.RoutingModel, routing: pywrapcp.RoutingModel, data_model: DataModel):
def distance_callback(from_index, to_index):
"""Returns the distance between the two nodes."""
# Convert from routing variable Index to distance matrix NodeIndex.
from_node = manager.IndexToNode(from_index)
to_node = manager.IndexToNode(to_index)
return data_model.distance_matrix[from_node][to_node]
transit_callback_index = routing.RegisterTransitCallback(distance_callback)
routing.SetArcCostEvaluatorOfAllVehicles(transit_callback_index)
dimension_name = 'Distance'
routing.AddDimension(
transit_callback_index,
0, # no slack
3000, # vehicle maximum travel distance
True, # start cumul to zero
dimension_name)
distance_dimension = routing.GetDimensionOrDie(dimension_name)
distance_dimension.SetGlobalSpanCostCoefficient(100)
for i in range(data_model.number_of_engineers):
routing.AddVariableMinimizedByFinalizer(
distance_dimension.CumulVar(routing.Start(i)))
routing.AddVariableMinimizedByFinalizer(
distance_dimension.CumulVar(routing.End(i)))
return routing
def print_solution(data: dict, routing: pywrapcp.RoutingModel,
assignment: pywrapcp.Assignment):
"""
Print routes on console.
"""
capacity_dimension = routing.GetDimensionOrDie('Capacity')
time_dimension = routing.GetDimensionOrDie('Time')
total_time = 0
d = {}
r = []
for vehicle_id in range(data['num_vehicles']):
index = routing.Start(vehicle_id)
node_props = []
while not routing.IsEnd(index):
props = node_properties(routing, assignment, capacity_dimension,
time_dimension, index)
node_props.append(props)
index = assignment.Value(routing.NextVar(index))
props = node_properties(routing, assignment, capacity_dimension,
time_dimension, index)
node_props.append(props)
route_time = assignment.Value(time_dimension.CumulVar(index))
route = "\n -> ".join(['[Node %2s: Load(%s) Time(%2s, %s)]' % prop \
for prop in node_props])
plan_output = f'Route for vehicle {vehicle_id}:\n {route}\n' + \
f'Load of the route: {props[1]}\nTime of the route: {route_time} min\n'
# print(plan_output)
#convert to dict
ddd = []
for node in node_props:
dd = {"node": node[0], "load": node[1], "time": [node[2], node[3]]}
ddd.append(dd)
r.append({
"vehicle_id": vehicle_id,
"route": ddd,
"total_time": route_time,
"load": len(ddd) - 2
})
total_time += route_time
d['solver'] = r
d['total_time'] = total_time
# print(f'Total time of all routes: {total_time} min')
return d
def apply(self, manager: pywrapcp.RoutingIndexManager,
routing: pywrapcp.RoutingModel, data_model: DataModel):
nodes = data_model.nodes
engineers = data_model.engineers_to_plan
for index in range(0, data_model.number_of_nodes):
if index < data_model.number_of_engineers:
routing.SetAllowedVehiclesForIndex([index], index)
else:
allowed_vehicles = [
ind for ind, node in enumerate(engineers)
if self.can_visit(data_model, nodes[index], engineers[ind])
]
routing.SetAllowedVehiclesForIndex(allowed_vehicles, index)
return routing
def _set_objective_function(problem: Problem, manager: RoutingIndexManager,
solver: RoutingModel):
"""Method to set the objective function of the Optimization Model"""
def _time_callback(from_index: int, to_index: int):
"""Callback to obtain the complete time between Stops"""
origin = manager.IndexToNode(from_index)
destination = manager.IndexToNode(to_index)
travelling_time = problem.estimations[(origin, destination)]
service_time = problem.stops[destination].service_time
return travelling_time + service_time
callback_index = solver.RegisterTransitCallback(_time_callback)
solver.SetArcCostEvaluatorOfAllVehicles(callback_index)
logging.info('Set the objective function to the OptimizationModel.')
def test_apply_constraints_capacity_constraint(self):
"""Asserts constraints are read correctly by the solver"""
model_builder = OptimizationModelBuilder(
constraints=[CapacityConstraint()])
problem = self.problem
manager = RoutingIndexManager(
len(problem.stops), # Number of locations
len(problem.vehicles), # Number of vehicles
problem.starts, # Start list of Vehicles
problem.ends # End list of Vehicles
)
solver = RoutingModel(manager)
model_builder._apply_constraints(problem, manager, solver)
self.assertTrue(solver, msg='Constraints added incorrectly.')
self.assertTrue(solver.HasDimension('capacity_constraint'),
msg='Capacity constraint not added.')
def apply(self, manager: pywrapcp.RoutingModel,
routing: pywrapcp.RoutingModel, data_model: DataModel):
def demand_callback(from_index):
"""Returns the demand of the node."""
# Convert from routing variable Index to demands NodeIndex.
from_node = manager.IndexToNode(from_index)
return data_model.demands[from_node]
demand_callback_index = routing.RegisterUnaryTransitCallback(
demand_callback)
routing.AddDimensionWithVehicleCapacity(
demand_callback_index,
0, # null capacity slack
data_model.vehicle_capacities,
True, # start cumul to zero
'Capacity')
return routing
def _set_balance_constraints_on_distance_driven(
self, data: VehicleRoutingProblemInstance, routing: RoutingModel,
deliveries: List[Delivery]):
# ========= BALANCE DISTRIBUTION CONSTRAIN =========
# https://activimetrics.com/blog/ortools/counting_dimension/
# A “fair” distribution of loads
average_load = int(
2 * len(deliveries) //
data.num_plans_to_create) # each delivery has 2 locations
count_dimension = routing.GetDimensionOrDie(
self.DISTANCE_DIMENSION_NAME)
for veh in range(0, data.num_plans_to_create):
index_end = routing.End(veh)
# https://github.com/google/or-tools/blob/v8.0/ortools/constraint_solver/routing.h#L2460
# 5 min penalty for every order over average load
count_dimension.SetCumulVarSoftLowerBound(index_end, average_load,
5 * 60)
def main() -> None:
"""
Entry point of the program.
"""
# Parse command line arguments
args = parse_args()
# Instantiate the data problem
data = load_data_model(args.path)
# Create the Routing Index Manager and Routing Model
manager = RoutingIndexManager(data["num_locations"], data["num_vehicles"],
data["depot"])
routing = RoutingModel(manager)
# Define weights of edges
weight_callback_index = routing.RegisterTransitCallback(
create_weight_callback(manager, data))
routing.SetArcCostEvaluatorOfAllVehicles(weight_callback_index)
# Add capacity constraints
demand_callback = create_demand_callback(manager, data)
demand_callback_index = routing.RegisterUnaryTransitCallback(
demand_callback)
add_capacity_constraints(routing, manager, data, demand_callback_index)
# Add time window constraints
time_callback_index = routing.RegisterTransitCallback(
create_time_callback(manager, data))
add_time_window_constraints(routing, manager, data, time_callback_index)
# Set first solution heuristic (cheapest addition)
search_params = DefaultRoutingSearchParameters()
search_params.first_solution_strategy = FirstSolutionStrategy.PATH_CHEAPEST_ARC
if args.gls:
search_params.local_search_metaheuristic = (
LocalSearchMetaheuristic.GUIDED_LOCAL_SEARCH)
# NOTE: Since Guided Local Search could take a very long time, we set a
# reasonable time limit
search_params.time_limit.seconds = 30
if args.verbose:
search_params.log_search = True
# Solve the problem
assignment = routing.SolveWithParameters(search_params)
if not assignment:
print("No solution found.")
return
# Print the solution
print_solution(data, routing, manager, assignment)
# Export network and route graphs
if args.export_network_graph:
draw_network_graph(args.export_network_graph, data)
if args.export_route_graph:
draw_route_graph(args.export_route_graph, data, routing, manager,
assignment)
def _extract_solution(self, assignment: Assignment, vrp_instance: VehicleRoutingProblemInstance,
manager: RoutingIndexManager, routing: RoutingModel) \
-> VehicleRoutingProblemSolution:
routes = []
etas = []
etds = []
time_dimension = routing.GetDimensionOrDie(
self.DURATION_DIMENSION_NAME)
for vehicle_id in range(vrp_instance.num_plans_to_create):
index = routing.Start(vehicle_id)
route = []
route_etas = []
route_etds = []
while not routing.IsEnd(index):
node = manager.IndexToNode(index)
if node in vrp_instance.pickup_nodes:
this_event_time = vrp_instance.pickup_service_time
elif node in vrp_instance.drop_nodes:
this_event_time = vrp_instance.drop_service_time
else:
this_event_time = 0
time_var = time_dimension.CumulVar(index)
eta = assignment.Min(time_var)
etd = assignment.Max(time_var) + this_event_time
route.append(node)
route_etas.append(eta)
route_etds.append(etd)
index = assignment.Value(routing.NextVar(index))
routes.append(route)
etas.append(route_etas)
etds.append(route_etds)
ret = VehicleRoutingProblemSolution(plans=routes, etas=etas, etds=etds)
return ret
def _set_balance_constraints_on_number_of_packages(
self, data: VehicleRoutingProblemInstance, routing: RoutingModel):
# ========= BALANCE DISTRIBUTION CONSTRAIN =========
routing.AddConstantDimension(
1, # increment by one every time
100000, # max value forces equivalent # of jobs
True, # set count to zero
self.COUNT_DIMENSION_NAME)
average_load = int(
2 * len(data.drop_nodes) //
data.num_plans_to_create) # each delivery has 2 locations
count_dimension = routing.GetDimensionOrDie(self.COUNT_DIMENSION_NAME)
for veh in range(0, data.num_plans_to_create):
index_end = routing.End(veh)
# https://github.com/google/or-tools/blob/v8.0/ortools/constraint_solver/routing.h#L2460
# 5 min penalty for every order over average load
count_dimension.SetCumulVarSoftUpperBound(index_end,
average_load + 6, 5 * 60)
def add_time_window_constraints(routing: pywrapcp.RoutingModel, data: dict,
time_callback):
"""
Add time window constraints.
"""
time = 'Time'
horizon = 120
routing.AddDimension(
evaluator=time_callback,
slack_max=horizon, # allow waiting time
capacity=horizon, # maximum time per vehicle
# Don't force start cumul to zero. This doesn't have any effect in this example,
# since the depot has a start window of (0, 0).
fix_start_cumul_to_zero=False,
name=time,
)
time_dimension = routing.GetDimensionOrDie(time)
for loc_node, (open_time, close_time) in enumerate(data['time_windows']):
index = routing.NodeToIndex(loc_node)
time_dimension.CumulVar(index).SetRange(open_time, close_time)
def _set_pickup_delivery_constraints(self,
data: VehicleRoutingProblemInstance,
routing: RoutingModel,
manager: RoutingIndexManager):
time_dimension = routing.GetDimensionOrDie(
self.DURATION_DIMENSION_NAME)
# ========= PICKUP/DELIVERY CONSTRAIN =========
# Define Transportation Requests.
solver: Solver = routing.solver()
for pickup_node, delivery_node in data.deliveries_not_started:
pickup_index = manager.NodeToIndex(pickup_node)
delivery_index = manager.NodeToIndex(delivery_node)
routing.AddPickupAndDelivery(pickup_index, delivery_index)
solver.Add(
routing.VehicleVar(pickup_index) == routing.VehicleVar(
delivery_index))
solver.Add(
time_dimension.CumulVar(pickup_index) <=
time_dimension.CumulVar(delivery_index))
# Define constraint of deliveries in progress - only vehicle that picked the order is able to finish it
for courier_idx, node in data.deliveries_in_progress:
index = manager.NodeToIndex(node)
routing.SetAllowedVehiclesForIndex([courier_idx], index)
def add_capacity_constraints(routing: pywrapcp.RoutingModel, data: dict,
demand_callback):
"""
Add capacity constraints.
"""
routing.AddDimensionWithVehicleCapacity(
evaluator=demand_callback,
slack_max=0, # null slack
# vehicle maximum capacities
vehicle_capacities=data['vehicle_capacities'],
fix_start_cumul_to_zero=True, # start cumul to zero
name='Capacity',
)
def add_time_window_constraints(
routing: RoutingModel,
manager: RoutingIndexManager,
data: dict,
time_callback_index: int,
) -> None:
max_ = 120
routing.AddDimension(
time_callback_index,
slack_max=
max_, # An upper bound for slack (the wait times at the locations)
capacity=
max_, # An upper bound for the total time over each vehicle's route
# Don't force start cumul to zero. This doesn't have any effect in this example,
# since the depot has a start window of (0, 0).
fix_start_cumul_to_zero=False,
name='Time',
)
time_dimension = routing.GetDimensionOrDie('Time')
for loc_idx, (open_time, close_time) in enumerate(data['time_windows']):
index = manager.NodeToIndex(loc_idx)
time_dimension.CumulVar(index).SetRange(open_time, close_time)
def _set_arc_durations(self, data: VehicleRoutingProblemInstance,
manager: RoutingIndexManager, routing: RoutingModel,
dimension_name: str):
# ========= DURATION CONSTRAIN =========
cost_callback_indices = []
# model "cost between nodes per vehicle"
for vehicle_idx in range(data.num_plans_to_create):
def vehicle_cost_callback(from_index, to_index):
from_node = manager.IndexToNode(from_index)
to_node = manager.IndexToNode(to_index)
time = data.car_duration_matrix[from_node][
to_node] # in seconds
is_pickup = from_node in data.pickup_nodes
is_drop = from_node in data.drop_nodes
waiting_time = 0
if is_pickup:
waiting_time = ConfigProvider.get_config(
).pickup_waiting_time
elif is_drop:
waiting_time = ConfigProvider.get_config(
).drop_waiting_time
return time + waiting_time
cost_callback_index = routing.RegisterTransitCallback(
vehicle_cost_callback)
cost_callback_indices.append(cost_callback_index)
routing.AddDimensionWithVehicleTransits(
cost_callback_indices,
MAX_TIMESTAMP_VALUE, # waiting time
MAX_TIMESTAMP_VALUE, # maximum time per vehicle
# since we are using timestamps as measuring unit we should not overflow
False, # Don't force start cumul to zero.
dimension_name)
def _extract_solution(self, manager: RoutingIndexManager,
routing: RoutingModel, assignment: Assignment,
indices_to_visit: List[int]) -> Dict[str, Any]:
"""Transform results to a usable format
Parameters
----------
manager : RoutingIndexManager
OR-tools' object to manage conversion between NodeIndex and variable index
routing : RoutingModel
OR-tools' object for route solving
assignment : Assignment
OR-tools' object for mapping from variable to domains
indices_to_visit : List[int]
The list of indices corresponding to the desired facilities to visit
Returns
-------
Dict[str, Any]
With keys:
- 'objective', set to objective value (minified distance)
- 'order', instructions for the order of facilities to visit
"""
sln = {"objective": assignment.ObjectiveValue()}
stop_indices = []
index = routing.Start(0)
while not routing.IsEnd(index):
relative_index = manager.IndexToNode(index)
stop_indices.append(indices_to_visit[relative_index])
previous_index = index
index = assignment.Value(routing.NextVar(index))
relative_index = manager.IndexToNode(index)
stop_indices.append(indices_to_visit[relative_index])
sln["order"] = stop_indices
return sln
def test_set_objective_function(self):
"""Asserts the objective function is added to the solver"""
model_builder = OptimizationModelBuilder(
constraints=[CapacityConstraint()])
problem = self.problem
manager = RoutingIndexManager(
len(problem.stops), # Number of locations
len(problem.vehicles), # Number of vehicles
problem.starts, # Start list of Vehicles
problem.ends # End list of Vehicles
)
solver = RoutingModel(manager)
model_builder._set_objective_function(problem, manager, solver)
self.assertTrue(solver, msg='Objective function set incorrectly.')
def add_time_window_constraints(
routing: RoutingModel,
manager: RoutingIndexManager,
data: dict,
time_callback_index: int,
) -> None:
"""
Add time window constraints.
"""
horizon = 120
routing.AddDimension(
time_callback_index,
slack_max=horizon, # allow waiting time
capacity=horizon, # maximum time per vehicle
# Don't force start cumul to zero. This doesn't have any effect in this example,
# since the depot has a start window of (0, 0).
fix_start_cumul_to_zero=False,
name='Time',
)
time_dimension = routing.GetDimensionOrDie('Time')
for loc_idx, (open_time, close_time) in enumerate(data['time_windows']):
index = manager.NodeToIndex(loc_idx)
time_dimension.CumulVar(index).SetRange(open_time, close_time)
def node_properties(
routing: pywrapcp.RoutingModel,
assignment: pywrapcp.Assignment,
capacity_dimension: pywrapcp.RoutingDimension,
time_dimension: pywrapcp.RoutingDimension,
index: int,
) -> tuple:
"""
Get a node's properties on the index.
"""
node_index = routing.IndexToNode(index)
load = assignment.Value(capacity_dimension.CumulVar(index))
time_var = time_dimension.CumulVar(index)
time_min, time_max = assignment.Min(time_var), assignment.Max(time_var)
return (node_index, load, time_min, time_max)
