def test_number_of_agents(self):
# World Definition
params = ParameterServer()
world = World(params)
# Model Definitions
behavior_model = BehaviorConstantAcceleration(params)
execution_model = ExecutionModelInterpolate(params)
dynamic_model = SingleTrackModel(params)
behavior_model2 = BehaviorConstantAcceleration(params)
execution_model2 = ExecutionModelInterpolate(params)
dynamic_model2 = SingleTrackModel(params)
# Map Definition
map_interface = MapInterface()
xodr_map = MakeXodrMapOneRoadTwoLanes()
map_interface.SetOpenDriveMap(xodr_map)
world.SetMap(map_interface)
agent_2d_shape = CarLimousine()
init_state = np.array([0, 13, -1.75, 0, 5])
agent_params = params.AddChild("agent1")
goal_polygon = Polygon2d(
[1, 1, 0],
[Point2d(0, 0),
Point2d(0, 2),
Point2d(2, 2),
Point2d(2, 0)])
goal_polygon = goal_polygon.Translate(Point2d(50, -2))
agent = Agent(init_state, behavior_model, dynamic_model,
execution_model, agent_2d_shape, agent_params,
GoalDefinitionPolygon(goal_polygon), map_interface)
world.AddAgent(agent)
init_state2 = np.array([0, 16, -1.75, 0, 5])
agent2 = Agent(init_state2, behavior_model2, dynamic_model2,
execution_model2, agent_2d_shape, agent_params,
GoalDefinitionPolygon(goal_polygon), map_interface)
world.AddAgent(agent2)
evaluator = EvaluatorNumberOfAgents(agent.id)
world.AddEvaluator("num_agents", evaluator)
info = world.Evaluate()
self.assertEqual(info["num_agents"], len(world.agents))
# do it once more
self.assertEqual(info["num_agents"], len(world.agents))
world.RemoveAgentById(agent2.id)
info = world.Evaluate()
# evaluator should still hold two
self.assertNotEqual(info["num_agents"], len(world.agents))
self.assertEqual(info["num_agents"], 2)
world.Step(0.1)
info = world.Evaluate()
# evaluator should still hold two
self.assertEqual(info["num_agents"], 2)
def test_gap_distance_front(self):
# World Definition
params = ParameterServer()
world = World(params)
gap = 10
# Model Definitions
behavior_model = BehaviorConstantAcceleration(params)
execution_model = ExecutionModelInterpolate(params)
dynamic_model = SingleTrackModel(params)
behavior_model2 = BehaviorConstantAcceleration(params)
execution_model2 = ExecutionModelInterpolate(params)
dynamic_model2 = SingleTrackModel(params)
# Map Definition
map_interface = MapInterface()
xodr_map = MakeXodrMapOneRoadTwoLanes()
map_interface.SetOpenDriveMap(xodr_map)
world.SetMap(map_interface)
agent_2d_shape = CarLimousine()
init_state = np.array([0, 13, -1.75, 0, 5])
agent_params = params.AddChild("agent1")
goal_polygon = Polygon2d(
[1, 1, 0],
[Point2d(0, 0),
Point2d(0, 2),
Point2d(2, 2),
Point2d(2, 0)])
goal_polygon = goal_polygon.Translate(Point2d(50, -2))
agent = Agent(init_state, behavior_model, dynamic_model,
execution_model, agent_2d_shape, agent_params,
GoalDefinitionPolygon(goal_polygon), map_interface)
world.AddAgent(agent)
init_state2 = np.array([0, 13 + gap, -1.75, 0, 5])
agent2 = Agent(init_state2, behavior_model2, dynamic_model2,
execution_model2, agent_2d_shape, agent_params,
GoalDefinitionPolygon(goal_polygon), map_interface)
world.AddAgent(agent2)
world.Step(0.1)
evaluator = EvaluatorGapDistanceFront(agent.id)
world.AddEvaluator("gap", evaluator)
info = world.Evaluate()
self.assertAlmostEqual(info["gap"],
gap - agent_2d_shape.front_dist -
agent_2d_shape.rear_dist,
places=4)
def test_database_runner(self):
dbs = DatabaseSerializer(test_scenarios=4, test_world_steps=5, num_serialize_scenarios=2)
dbs.process("data/database1")
local_release_filename = dbs.release(version="test")
db = BenchmarkDatabase(database_root=local_release_filename)
evaluators = {"success" : "EvaluatorGoalReached", "collision" : "EvaluatorCollisionEgoAgent",
"max_steps": "EvaluatorStepCount"}
terminal_when = {"collision" :lambda x: x, "max_steps": lambda x : x>2}
params = ParameterServer() # only for evaluated agents not passed to scenario!
behaviors_tested = {"IDM": BehaviorIDMClassic(params), "Const" : BehaviorConstantAcceleration(params)}
benchmark_runner = BenchmarkRunner(benchmark_database=db,
evaluators=evaluators,
terminal_when=terminal_when,
behaviors=behaviors_tested,
log_eval_avg_every=5)
result = benchmark_runner.run()
df = result.get_data_frame()
print(df)
self.assertEqual(len(df.index), 2*2*2) # 2 Behaviors * 2 Serialize Scenarios * 1 scenario sets
groups = result.get_evaluation_groups()
self.assertEqual(set(groups), set(["behavior", "scen_set"]))
def test_database_multiprocessing_history(self):
dbs = DatabaseSerializer(test_scenarios=4, test_world_steps=5, num_serialize_scenarios=2)
dbs.process("data/database1")
local_release_filename = dbs.release(version="test")
db = BenchmarkDatabase(database_root=local_release_filename)
evaluators = {"success" : "EvaluatorGoalReached", "collision" : "EvaluatorCollisionEgoAgent",
"max_steps": "EvaluatorStepCount"}
terminal_when = {"collision" :lambda x: x, "max_steps": lambda x : x>2}
params = ParameterServer() # only for evaluated agents not passed to scenario!
behaviors_tested = {"IDM": BehaviorIDMClassic(params), "Const" : BehaviorConstantAcceleration(params)}
benchmark_runner = BenchmarkRunnerMP(benchmark_database=db,
evaluators=evaluators,
terminal_when=terminal_when,
behaviors=behaviors_tested,
log_eval_avg_every=10)
rst = benchmark_runner.run(maintain_history=True)
self.assertEqual(len(rst.get_histories()), 2*2*2)
rst = benchmark_runner.run_benchmark_config(3, viewer=None, maintain_history=True)
scenario_history = rst.get_histories()[3]
print(scenario_history)
params = ParameterServer()
viewer = MPViewer(
params=params,
x_range=[5060, 5160],
y_range=[5070,5150],
use_world_bounds=True)
viewer.drawWorld(world=scenario_history[1].GetWorldState(),
eval_agent_ids=scenario_history[1].eval_agent_ids)
viewer.show(block=True)
def test_database_multiprocessing_runner(self):
dbs = DatabaseSerializer(test_scenarios=4, test_world_steps=5, num_serialize_scenarios=5)
dbs.process("data/database1")
local_release_filename = dbs.release(version="test")
db = BenchmarkDatabase(database_root=local_release_filename)
evaluators = {"success" : "EvaluatorGoalReached", "collision" : "EvaluatorCollisionEgoAgent",
"max_steps": "EvaluatorStepCount"}
terminal_when = {"collision" :lambda x: x, "max_steps": lambda x : x>2}
params = ParameterServer() # only for evaluated agents not passed to scenario!
behaviors_tested = {"IDM": BehaviorIDMClassic(params), "Const" : BehaviorConstantAcceleration(params)}
benchmark_runner = BenchmarkRunnerMP(benchmark_database=db,
evaluators=evaluators,
terminal_when=terminal_when,
behaviors=behaviors_tested,
log_eval_avg_every=10)
result = benchmark_runner.run()
df = result.get_data_frame()
print(df)
self.assertEqual(len(df.index), 20) # 2 Behaviors * 5 Serialize Scenarios * 2 scenario sets
params2 = ParameterServer()
viewer = MPViewer(
params=params2,
x_range=[5060, 5160],
y_range=[5070,5150],
use_world_bounds=True)
rst = benchmark_runner.run_benchmark_config(10, viewer=viewer)
def test_planning_time(self):
param_server = ParameterServer()
# Model Definition
behavior_model = BehaviorConstantAcceleration(param_server)
execution_model = ExecutionModelInterpolate(param_server)
dynamic_model = SingleTrackModel(param_server)
# Agent Definition
agent_2d_shape = CarLimousine()
init_state = np.array([0, -191.789,-50.1725, 3.14*3.0/4.0, 150/3.6])
agent_params = param_server.AddChild("agent1")
goal_polygon = Polygon2d([0, 0, 0],
[Point2d(-4,-4),
Point2d(-4,4),
Point2d(4,4),
Point2d(4,-4)])
goal_polygon = goal_polygon.Translate(Point2d(-191.789,-50.1725))
agent = Agent(init_state,
behavior_model,
dynamic_model,
execution_model,
agent_2d_shape,
agent_params,
GoalDefinitionPolygon(goal_polygon),
None)
world = World(param_server)
world.AddAgent(agent)
evaluator = EvaluatorPlanningTime(agent.id)
world.AddEvaluator("time", evaluator)
info = world.Evaluate()
self.assertEqual(info["time"], 0.0)
def test_one_agent_at_goal_state_limits(self):
param_server = ParameterServer()
# Model Definition
behavior_model = BehaviorConstantAcceleration(param_server)
execution_model = ExecutionModelInterpolate(param_server)
dynamic_model = SingleTrackModel(param_server)
# Agent Definition
agent_2d_shape = CarLimousine()
init_state = np.array([0, -191.789,-50.1725, 3.14*3.0/4.0, 150/3.6])
agent_params = param_server.AddChild("agent1")
goal_polygon = Polygon2d([0, 0, 0],
[Point2d(-1,-1),
Point2d(-1,1),
Point2d(1,1),
Point2d(1,-1)])
goal_polygon = goal_polygon.Translate(Point2d(-191.789,-50.1725))
agent = Agent(init_state,
behavior_model,
dynamic_model,
execution_model,
agent_2d_shape,
agent_params,
GoalDefinitionStateLimits(goal_polygon, (3.14*3.0/4.0-0.08, 3.14*3.0/4.0+0.08)),
None)
world = World(param_server)
world.AddAgent(agent)
evaluator = EvaluatorGoalReached(agent.id)
world.AddEvaluator("success", evaluator)
info = world.Evaluate()
self.assertEqual(info["success"], True)
def test_database_runner(self):
dbs = DatabaseSerializer(test_scenarios=4,
test_world_steps=5,
num_serialize_scenarios=2)
dbs.process("data/database1")
local_release_filename = dbs.release(version="test")
db = BenchmarkDatabase(database_root=local_release_filename)
safe_dist_params = ParameterServer(log_if_default=True)
test_python_params = ParameterServer(log_if_default=True)
evaluators = {
"success": "EvaluatorGoalReached",
"collision": "EvaluatorCollisionEgoAgent",
"max_steps": "EvaluatorStepCount",
"safe_dist_lon": {
"type": "EvaluatorDynamicSafeDist",
"params": safe_dist_params
},
"safe_dist_lat": {
"type": "EvaluatorStaticSafeDist",
"params": safe_dist_params
},
"test_evaluator": {
"type": "TestPythonEvaluator",
"params": test_python_params
},
"test_evaluator_serializable": TestPythonEvaluatorSerializable()
}
terminal_when = {
"collision": lambda x: x,
"max_steps": lambda x: x > 2,
"safe_dist_lon": lambda x: x
}
params = ParameterServer(
) # only for evaluated agents not passed to scenario!
behaviors_tested = {
"IDM": BehaviorIDMClassic(params),
"Const": BehaviorConstantAcceleration(params)
}
EvaluationConfig.AddEvaluationModule(
"bark.benchmark.tests.test_evaluator")
benchmark_runner = BenchmarkRunner(benchmark_database=db,
evaluators=evaluators,
log_eval_avg_every=1,
terminal_when=terminal_when,
behaviors=behaviors_tested)
result = benchmark_runner.run()
df = result.get_data_frame()
print(df)
self.assertEqual(
len(df.index),
2 * 2 * 2) # 2 Behaviors * 2 Serialize Scenarios * 1 scenario sets
result.load_benchmark_configs()
groups = result.get_evaluation_groups()
self.assertEqual(set(groups), set(["behavior", "scen_set"]))
def test_database_runner_checkpoint(self):
dbs = DatabaseSerializer(test_scenarios=4, test_world_steps=5, num_serialize_scenarios=10)
dbs.process("data/database1")
local_release_filename = dbs.release(version="test")
db = BenchmarkDatabase(database_root=local_release_filename)
evaluators = {"success" : "EvaluatorGoalReached", "collision" : "EvaluatorCollisionEgoAgent",
"max_steps": "EvaluatorStepCount"}
terminal_when = {"collision" :lambda x: x, "max_steps": lambda x : x>2}
params = ParameterServer() # only for evaluated agents not passed to scenario!
behaviors_tested = {"IDM": BehaviorIDMClassic(params), "Const" : BehaviorConstantAcceleration(params)}
benchmark_runner = BenchmarkRunner(benchmark_database=db,
evaluators=evaluators,
terminal_when=terminal_when,
behaviors=behaviors_tested,
log_eval_avg_every=20,
checkpoint_dir="checkpoints1/")
# one run after 30 steps benchmark dumped
result = benchmark_runner.run(checkpoint_every = 30)
df = result.get_data_frame()
print(df)
self.assertEqual(len(df.index), 40) # 2 Behaviors * 10 Serialize Scenarios * 2 scenario sets
# check twice first, merging from checkpoints
merged_result = BenchmarkRunner.merge_checkpoint_benchmark_results(checkpoint_dir="checkpoints1/")
df = merged_result.get_data_frame()
self.assertEqual(len(df.index), 30)
# second load merged results
self.assertTrue(os.path.exists(os.path.join("checkpoints1/merged_results.ckpnt")))
merged_result = BenchmarkRunner.merge_checkpoint_benchmark_results(checkpoint_dir="checkpoints1/")
df = merged_result.get_data_frame()
self.assertEqual(len(df.index), 30)
configs_to_run = BenchmarkRunner.get_configs_to_run(benchmark_runner.configs_to_run, merged_result)
self.assertEqual(len(configs_to_run), 10)
benchmark_runner2 = BenchmarkRunner(benchmark_database=db,
evaluators=evaluators,
terminal_when=terminal_when,
behaviors=behaviors_tested,
log_eval_avg_every=1,
checkpoint_dir="checkpoints1/",
merge_existing=True)
result = benchmark_runner2.run(checkpoint_every = 7)
df = result.get_data_frame()
print(df)
self.assertEqual(len(df.index), 40) # 2 Behaviors * 10 Serialize Scenarios * 2 scenario sets
# check if results maintained in existing result dump, 30 from previous run + 7 after new checkpoint
merged_result = BenchmarkRunner.merge_checkpoint_benchmark_results(checkpoint_dir="checkpoints1/")
df = merged_result.get_data_frame()
self.assertEqual(len(df.index), 37)
def test_database_multiprocessing_runner_checkpoint(self):
dbs = DatabaseSerializer(test_scenarios=1, test_world_steps=2, num_serialize_scenarios=10)
dbs.process("data/database1")
local_release_filename = dbs.release(version="test")
db = BenchmarkDatabase(database_root=local_release_filename)
evaluators = {"success" : "EvaluatorGoalReached", "collision" : "EvaluatorCollisionEgoAgent",
"max_steps": "EvaluatorStepCount"}
terminal_when = {"collision" :lambda x: x, "max_steps": lambda x : x>2}
params = ParameterServer() # only for evaluated agents not passed to scenario!
behaviors_tested = {"IDM": BehaviorIDMClassic(params), "Const" : BehaviorConstantAcceleration(params)}
benchmark_runner = BenchmarkRunnerMP(benchmark_database=db,
evaluators=evaluators,
terminal_when=terminal_when,
behaviors=behaviors_tested,
log_eval_avg_every=10,
num_cpus=4,
checkpoint_dir="checkpoints2/",
merge_existing=False)
benchmark_runner.clear_checkpoint_dir()
# one run after 30 steps benchmark dumped
result = benchmark_runner.run(checkpoint_every = 3)
df = result.get_data_frame()
print(df)
self.assertEqual(len(df.index), 40) # 2 Behaviors * 10 Serialize Scenarios * 2 scenario sets
merged_result = BenchmarkRunner.merge_checkpoint_benchmark_results(checkpoint_dir="checkpoints2/")
df = merged_result.get_data_frame()
self.assertEqual(len(df.index), 4*9)
# self.assertEqual(len(merged_result.get_histories()), 4*9)
self.assertEqual(len(merged_result.get_benchmark_configs()), 4*9)
configs_to_run = BenchmarkRunner.get_configs_to_run(benchmark_runner.configs_to_run, merged_result)
self.assertEqual(len(configs_to_run), 4)
ray.shutdown()
benchmark_runner2 = BenchmarkRunnerMP(benchmark_database=db,
evaluators=evaluators,
terminal_when=terminal_when,
behaviors=behaviors_tested,
log_eval_avg_every=1,
checkpoint_dir="checkpoints2/",
merge_existing=True)
result = benchmark_runner2.run(checkpoint_every = 1)
df = result.get_data_frame()
print(df)
self.assertEqual(len(df.index), 40) # 2 Behaviors * 10 Serialize Scenarios * 2 scenario sets
# check if existing result is incorporated for mergin result
merged_result = BenchmarkRunner.merge_checkpoint_benchmark_results(checkpoint_dir="checkpoints2/")
df = merged_result.get_data_frame()
self.assertEqual(len(df.index), 40)
def __init__(self, init_state, goal_polygon, map_interface, params):
behavior_model = BehaviorConstantAcceleration(params)
execution_model = ExecutionModelInterpolate(params)
dynamic_model = SingleTrackModel(params)
agent_2d_shape = CarLimousine()
agent_params = params.AddChild("agent")
super(TestAgent,
self).__init__(init_state, behavior_model, dynamic_model,
execution_model, agent_2d_shape, agent_params,
GoalDefinitionPolygon(goal_polygon),
map_interface)
def setUp(self):
param_server = ParameterServer()
world = World(param_server)
self.defaults = dict()
self.defaults["world"] = world
self.defaults["ego_behavior"] = BehaviorConstantAcceleration(
param_server)
self.defaults["ego_dynamic"] = SingleTrackModel(param_server)
self.defaults["ego_execution"] = ExecutionModelInterpolate(
param_server)
self.defaults["ego_shape"] = CarLimousine()
self.defaults["other_behavior"] = BehaviorConstantAcceleration(
param_server)
self.defaults["other_dynamic"] = SingleTrackModel(param_server)
self.defaults["other_execution"] = ExecutionModelInterpolate(
param_server)
self.defaults["other_shape"] = CarLimousine()
self.defaults["agent_params"] = param_server.addChild("agent")
self.defaults["default_vehicle_dynamics"] = [
1.7, -1.7, -1.69, -1.67, 0.2, -0.8, 0.1, 1.
]
def test_write_params_agent(self):
params = ParameterServer()
behavior = BehaviorConstantAcceleration(params)
execution = ExecutionModelInterpolate(params)
dynamic = SingleTrackModel(params)
shape = Polygon2d([1.25, 1, 0], [
Point2d(0, 0),
Point2d(0, 2),
Point2d(4, 2),
Point2d(4, 0),
Point2d(0, 0)
])
init_state = np.zeros(4)
agent = Agent(init_state, behavior, dynamic, execution, shape,
params.AddChild("agent"))
params.Save("written_agents_param_test.json")
def test_database_multiprocessing_runner(self):
dbs = DatabaseSerializer(test_scenarios=4,
test_world_steps=5,
num_serialize_scenarios=5)
dbs.process("data/database1")
local_release_filename = dbs.release(version="test")
db = BenchmarkDatabase(database_root=local_release_filename)
test_python_params = ParameterServer(log_if_default=True)
evaluators = {
"success": "EvaluatorGoalReached",
"collision": "EvaluatorCollisionEgoAgent",
"max_steps": "EvaluatorStepCount",
"test_evaluator": {
"type": "TestPythonEvaluator",
"params": test_python_params
}
}
terminal_when = {
"collision": lambda x: x,
"max_steps": lambda x: x > 2
}
params = ParameterServer(
) # only for evaluated agents not passed to scenario!
behaviors_tested = {
"IDM": BehaviorIDMClassic(params),
"Const": BehaviorConstantAcceleration(params)
}
EvaluationConfig.AddEvaluationModule(
"bark.benchmark.tests.test_evaluator")
benchmark_runner = BenchmarkRunnerMP(benchmark_database=db,
evaluators=evaluators,
terminal_when=terminal_when,
behaviors=behaviors_tested,
log_eval_avg_every=10)
benchmark_runner.clear_checkpoint_dir()
result = benchmark_runner.run()
df = result.get_data_frame()
print(df)
self.assertEqual(
len(df.index),
20) # 2 Behaviors * 5 Serialize Scenarios * 2 scenario sets
def test_database_runner_checkpoint(self):
dbs = DatabaseSerializer(test_scenarios=4,
test_world_steps=5,
num_serialize_scenarios=10)
dbs.process("data/database1")
local_release_filename = dbs.release(version="test")
db = BenchmarkDatabase(database_root=local_release_filename)
evaluators = {
"success": "EvaluatorGoalReached",
"collision": "EvaluatorCollisionEgoAgent",
"max_steps": "EvaluatorStepCount"
}
terminal_when = {
"collision": lambda x: x,
"max_steps": lambda x: x > 2
}
params = ParameterServer(
) # only for evaluated agents not passed to scenario!
behaviors_tested = {
"IDM": BehaviorIDMClassic(params),
"Const": BehaviorConstantAcceleration(params)
}
benchmark_runner = BenchmarkRunner(benchmark_database=db,
evaluators=evaluators,
terminal_when=terminal_when,
behaviors=behaviors_tested,
log_eval_avg_every=20,
checkpoint_dir="checkpoints1/")
benchmark_runner.clear_checkpoint_dir()
# one run after 30 steps benchmark dumped
result = benchmark_runner.run(checkpoint_every=30)
df = result.get_data_frame()
print(df)
self.assertEqual(
len(df.index),
40) # 2 Behaviors * 10 Serialize Scenarios * 2 scenario sets
# check twice first, merging from checkpoints
merged_result = BenchmarkRunner.merge_checkpoint_benchmark_results(
checkpoint_dir="checkpoints1/")
df = merged_result.get_data_frame()
self.assertEqual(len(df.index), 40)
def test_draw_agents(self):
params = ParameterServer()
behavior = BehaviorConstantAcceleration(params)
execution = ExecutionModelInterpolate(params)
dynamic = SingleTrackModel(params)
shape = Polygon2d([1.25, 1, 0], [
Point2d(0, 0),
Point2d(0, 2),
Point2d(4, 2),
Point2d(4, 0),
Point2d(0, 0)
])
shape2 = CarLimousine()
init_state = [0, 3, 2, 1]
init_state2 = [0, 0, 5, 4]
agent = Agent(init_state, behavior, dynamic, execution, shape,
params.AddChild("agent"))
agent2 = Agent(init_state2, behavior, dynamic, execution, shape2,
params.AddChild("agent"))
def test_world(self):
# create agent
params = ParameterServer()
behavior = BehaviorConstantAcceleration(params)
execution = ExecutionModelInterpolate(params)
dynamic = SingleTrackModel(params)
shape = Polygon2d([1.25, 1, 0], [
Point2d(0, 0),
Point2d(0, 2),
Point2d(4, 2),
Point2d(4, 0),
Point2d(0, 0)
])
init_state = np.array([0, 0, 0, 0, 5])
agent = Agent(init_state, behavior, dynamic, execution, shape,
params.AddChild("agent"))
road_map = OpenDriveMap()
newXodrRoad = XodrRoad()
newXodrRoad.id = 1
newXodrRoad.name = "Autobahn A9"
newPlanView = PlanView()
newPlanView.AddLine(Point2d(0, 0), 1.57079632679, 10)
newXodrRoad.plan_view = newPlanView
line = newXodrRoad.plan_view.GetReferenceLine().ToArray()
p = Point2d(line[-1][0], line[-1][1])
newXodrRoad.plan_view.AddSpiral(p, 1.57079632679, 50.0, 0.0, 0.3, 0.4)
line = newXodrRoad.plan_view.GetReferenceLine()
lane_section = XodrLaneSection(0)
lane = XodrLane()
lane.line = line
lane_section.AddLane(lane)
newXodrRoad.AddLaneSection(lane_section)
road_map.AddRoad(newXodrRoad)
r = Roadgraph()
map_interface = MapInterface()
map_interface.SetOpenDriveMap(road_map)
map_interface.SetRoadgraph(r)
world = World(params)
world.AddAgent(agent)
def test_relevent_agents(self):
map = "bark/runtime/tests/data/city_highway_straight.xodr"
map_interface = EvaluatorRSSTests.load_map(map)
world = self.defaults["world"].Copy()
world.SetMap(map_interface)
goal_polygon_1 = Polygon2d(
[0, 0, 0],
[Point2d(-1, -1),
Point2d(-1, 1),
Point2d(1, 1),
Point2d(1, -1)])
goal_polygon_1 = goal_polygon_1.Translate(Point2d(5.5, 120))
goal_polygon_2 = Polygon2d(
[0, 0, 0],
[Point2d(-1, -1),
Point2d(-1, 1),
Point2d(1, 1),
Point2d(1, -1)])
goal_polygon_2 = goal_polygon_2.Translate(Point2d(1.8, 120))
# Hard coded
ego_state = np.array([0, 5.5, 10, 0, 10])
other_1_state = np.array([0, 1.8, -10, 0, 15])
other_2_state = np.array([0, 1.8, -120, 0, 10])
ego = Agent(ego_state, self.defaults["ego_behavior"],
self.defaults["ego_dynamic"],
self.defaults["ego_execution"], self.defaults["ego_shape"],
self.defaults["agent_params"],
GoalDefinitionPolygon(goal_polygon_1), map_interface)
other_1 = Agent(other_1_state, self.defaults["other_behavior"],
self.defaults["other_dynamic"],
self.defaults["other_execution"],
self.defaults["other_shape"],
self.defaults["agent_params"],
GoalDefinitionPolygon(goal_polygon_2), map_interface)
other_2_behavior = BehaviorConstantAcceleration(
self.defaults["agent_params"])
other_2_dynamic = SingleTrackModel(self.defaults["agent_params"])
other_2_execution = ExecutionModelInterpolate(
self.defaults["agent_params"])
other_2 = Agent(other_2_state, other_2_behavior, other_2_dynamic,
other_2_execution, self.defaults["other_shape"],
self.defaults["agent_params"],
GoalDefinitionPolygon(goal_polygon_2), map_interface)
world.AddAgent(ego)
world.AddAgent(other_1)
world.AddAgent(other_2)
evaluator_rss = EvaluatorRss(ego.id, map,
self.defaults["default_vehicle_dynamics"])
world.Step(1)
responses = evaluator_rss.PairwiseEvaluate(world)
self.assertEqual(1, len(responses)) # Test GetRelevantAgents
self.assertTrue(responses[other_1.id])
self.assertFalse(other_2.id in responses)
def test_database_run_and_analyze(self):
dbs = DatabaseSerializer(test_scenarios=2, test_world_steps=3, num_serialize_scenarios=2)
# to find database files
cwd = os.getcwd()
if not debug:
os.chdir("../benchmark_database/")
else:
os.chdir("bazel-bin/bark/benchmark/tests/py_benchmark_process_tests.runfiles/benchmark_database")
dbs.process("data/database1")
local_release_filename = dbs.release(version="test")
db = BenchmarkDatabase(database_root=local_release_filename)
evaluators = {"success" : "EvaluatorGoalReached", "collision" : "EvaluatorCollisionEgoAgent",
"max_steps": "EvaluatorStepCount"}
terminal_when = {"collision" :lambda x: x, "max_steps": lambda x : x>2}
params = ParameterServer() # only for evaluated agents not passed to scenario!
behaviors_tested = {"IDM": BehaviorIDMClassic(params), "Const" : BehaviorConstantAcceleration(params)}
benchmark_runner = BenchmarkRunnerMP(benchmark_database=db,
evaluators=evaluators,
terminal_when=terminal_when,
behaviors=behaviors_tested,
log_eval_avg_every=2)
result = benchmark_runner.run(maintain_history=True)
result.dump(os.path.join("./benchmark_results"), dump_configs=True, \
dump_histories=True, max_mb_per_file=1)
result_loaded = BenchmarkResult.load(os.path.join("./benchmark_results"))
result_loaded.load_histories()
result_loaded.load_benchmark_configs()
params2 = ParameterServer()
fig = plt.figure(figsize=[10, 10])
viewer = MPViewer(
params=params2,
center=[5112, 5165],
y_length = 120,
enforce_y_length=True,
axis = fig.gca())
analyzer = BenchmarkAnalyzer(benchmark_result=result_loaded)
configs = analyzer.find_configs(criteria={"behavior": lambda x: x=="IDM", "success": lambda x : not x})
configs_const = analyzer.find_configs(criteria={"behavior": lambda x: x=="Const", "success": lambda x : not x})
os.chdir(cwd)
#analyzer.visualize(configs_idx_list = configs,
# viewer = viewer, real_time_factor=10, fontsize=12)
plt.close(fig)
fig, (ax1, ax2) = plt.subplots(1, 2)
viewer1 = MPViewer(
params=params2,
center=[5112, 5165],
y_length = 120,
enforce_y_length=True,
axis = ax1)
viewer2 = MPViewer(
params=params2,
center=[5112, 5165],
y_length = 120,
enforce_y_length=True,
axis = ax2)
analyzer.visualize(configs_idx_list = [configs[1:3], configs_const[1:3]],
viewer = [viewer1, viewer2], viewer_names=["IDM", "ConstVelocity"], real_time_factor=10, fontsize=12)
def run_configuration(argv):
params = ParameterServer()
# NOTE: Modify these paths to specify your preferred path for checkpoints and summaries
# params["ML"]["BehaviorTFAAgents"]["CheckpointPath"] = "/Users/hart/Development/bark-ml/checkpoints/"
# params["ML"]["TFARunner"]["SummaryPath"] = "/Users/hart/Development/bark-ml/checkpoints/"
params["Visualization"]["Agents"]["Alpha"]["Other"] = 0.2
params["Visualization"]["Agents"]["Alpha"]["Controlled"] = 0.2
params["Visualization"]["Agents"]["Alpha"]["Controlled"] = 0.2
params["ML"]["VisualizeCfWorlds"] = False
params["ML"]["VisualizeCfHeatmap"] = False
# params["ML"]["ResultsFolder"] = "/Users/hart/Development/bark-ml/results/data/"
# viewer = MPViewer(
# params=params,
# x_range=[-35, 35],
# y_range=[-35, 35],
# follow_agent_id=True)
# create environment
bp = ContinuousMergingBlueprint(params,
num_scenarios=2500,
random_seed=0)
observer = GraphObserver(params=params)
behavior_model_pool = []
for count, a in enumerate([-5., 0., 5.]):
local_params = params.AddChild("local_"+str(count))
local_params["BehaviorConstantAcceleration"]["ConstAcceleration"] = a
behavior = BehaviorConstantAcceleration(local_params)
behavior_model_pool.append(behavior)
env = CounterfactualRuntime(
blueprint=bp,
observer=observer,
render=False,
params=params,
behavior_model_pool=behavior_model_pool)
sac_agent = BehaviorGraphSACAgent(environment=env,
observer=observer,
params=params)
env.ml_behavior = sac_agent
runner = SACRunner(params=params,
environment=env,
agent=sac_agent)
if FLAGS.mode == "train":
runner.SetupSummaryWriter()
runner.Train()
elif FLAGS.mode == "visualize":
runner._environment._max_col_rate = 0.
runner.Run(num_episodes=1, render=True)
elif FLAGS.mode == "evaluate":
for cr in np.arange(0, 1, 0.1):
runner._environment._max_col_rate = cr
runner.Run(num_episodes=250, render=False, max_col_rate=cr)
runner._environment._tracer.Save(
params["ML"]["ResultsFolder"] + "evaluation_results_runtime.pckl")
goal_reached = runner._tracer.success_rate
runner._tracer.Save(
params["ML"]["ResultsFolder"] + "evaluation_results_runner.pckl")
def test_one_agent_at_goal_state_limits_frenet(self):
param_server = ParameterServer()
# Model Definition
behavior_model = BehaviorConstantAcceleration(param_server)
execution_model = ExecutionModelInterpolate(param_server)
dynamic_model = SingleTrackModel(param_server)
# Agent Definition
agent_2d_shape = CarLimousine()
agent_params = param_server.AddChild("agent1")
center_line = Line2d()
center_line.AddPoint(Point2d(5.0, 5.0))
center_line.AddPoint(Point2d(10.0, 10.0))
center_line.AddPoint(Point2d(20.0, 10.0))
max_lateral_dist = (0.4,1)
max_orientation_diff = (0.08, 0.1)
velocity_range = (20.0, 25.0)
goal_definition = GoalDefinitionStateLimitsFrenet(center_line,
max_lateral_dist, max_orientation_diff,
velocity_range)
# not at goal x,y, others yes
agent1 = Agent(np.array([0, 6, 8, 3.14/4.0 , velocity_range[0]]),
behavior_model,
dynamic_model,
execution_model,
agent_2d_shape,
agent_params,
goal_definition,
None)
# at goal x,y and others
agent2 = Agent(np.array([0, 5.0, 5.5, 3.14/4.0 , velocity_range[1]]),
behavior_model,
dynamic_model,
execution_model,
agent_2d_shape,
agent_params,
goal_definition,
None)
# not at goal x,y,v yes but not orientation
agent3 = Agent(np.array([0, 5, 5.5, 3.14/4.0+max_orientation_diff[1]+0.001 , 20]),
behavior_model,
dynamic_model,
execution_model,
agent_2d_shape,
agent_params,
goal_definition,
None)
# not at goal x,y, orientation but not v
agent4 = Agent(np.array([0, 5, 4.5, 3.14/4-max_orientation_diff[0], velocity_range[0]-0.01]),
behavior_model,
dynamic_model,
execution_model,
agent_2d_shape,
agent_params,
goal_definition,
None)
# at goal x,y, at lateral limit
agent5 = Agent(np.array([0, 15, 10-max_lateral_dist[0]+0.05, 0, velocity_range[1]]),
behavior_model,
dynamic_model,
execution_model,
agent_2d_shape,
agent_params,
goal_definition,
None)
# not at goal x,y slightly out of lateral limit
agent6 = Agent(np.array([0, 15, 10+max_lateral_dist[0]+0.05, 3.14/4+max_orientation_diff[0], velocity_range[0]]),
behavior_model,
dynamic_model,
execution_model,
agent_2d_shape,
agent_params,
goal_definition,
None)
# not at goal x,y,v yes but not orientation
agent7 = Agent(np.array([0, 5, 5.5, 3.14/4.0-max_orientation_diff[0]-0.001 , 20]),
behavior_model,
dynamic_model,
execution_model,
agent_2d_shape,
agent_params,
goal_definition,
None)
world = World(param_server)
world.AddAgent(agent1)
world.AddAgent(agent2)
world.AddAgent(agent3)
world.AddAgent(agent4)
world.AddAgent(agent5)
world.AddAgent(agent6)
world.AddAgent(agent7)
evaluator1 = EvaluatorGoalReached(agent1.id)
evaluator2 = EvaluatorGoalReached(agent2.id)
evaluator3 = EvaluatorGoalReached(agent3.id)
evaluator4 = EvaluatorGoalReached(agent4.id)
evaluator5 = EvaluatorGoalReached(agent5.id)
evaluator6 = EvaluatorGoalReached(agent6.id)
evaluator7 = EvaluatorGoalReached(agent7.id)
world.AddEvaluator("success1", evaluator1)
world.AddEvaluator("success2", evaluator2)
world.AddEvaluator("success3", evaluator3)
world.AddEvaluator("success4", evaluator4)
world.AddEvaluator("success5", evaluator5)
world.AddEvaluator("success6", evaluator6)
world.AddEvaluator("success7", evaluator7)
info = world.Evaluate()
self.assertEqual(info["success1"], False)
self.assertEqual(info["success2"], True)
self.assertEqual(info["success3"], False)
self.assertEqual(info["success4"], False)
self.assertEqual(info["success5"], True)
self.assertEqual(info["success6"], False)
self.assertEqual(info["success7"], False)
def test_evaluator_drivable_area(self):
# World Definition
params = ParameterServer()
world = World(params)
# Model Definitions
behavior_model = BehaviorConstantAcceleration(params)
execution_model = ExecutionModelInterpolate(params)
dynamic_model = SingleTrackModel(params)
# Map Definition
map_interface = MapInterface()
xodr_map = MakeXodrMapOneRoadTwoLanes()
map_interface.SetOpenDriveMap(xodr_map)
world.SetMap(map_interface)
#open_drive_map = world.map.GetOpenDriveMap()
#agent_2d_shape = CarLimousine()
agent_2d_shape = Polygon2d(
[1.25, 1, 0],
[Point2d(-1, -1),
Point2d(-1, 1),
Point2d(3, 1),
Point2d(3, -1)])
init_state = np.array([0, 3, -1.75, 0, 5])
agent_params = params.AddChild("agent1")
goal_polygon = Polygon2d(
[1, 1, 0],
[Point2d(0, 0),
Point2d(0, 2),
Point2d(2, 2),
Point2d(2, 0)])
goal_polygon = goal_polygon.Translate(Point2d(50, -2))
agent = Agent(
init_state,
behavior_model,
dynamic_model,
execution_model,
agent_2d_shape,
agent_params,
GoalDefinitionPolygon(goal_polygon), # goal_lane_id
map_interface)
world.AddAgent(agent)
evaluator = EvaluatorDrivableArea()
world.AddEvaluator("drivable_area", evaluator)
info = world.Evaluate()
self.assertFalse(info["drivable_area"])
viewer = MPViewer(params=params, use_world_bounds=True)
# Draw map
viewer.drawGoalDefinition(goal_polygon,
color=(1, 0, 0),
alpha=0.5,
facecolor=(1, 0, 0))
viewer.drawWorld(world)
viewer.drawRoadCorridor(agent.road_corridor)
viewer.show(block=False)
