features PhD #561
features PhD #561
Conversation
juloberno
commented
Dec 3, 2021
juloberno
commented
- benchmark runner: fix processing splitting into smaller files within zip file
- benchmark runner: scenario-type-specific terminal criteria
- Stochastic IDM with an intent mechanism (requires interface adaptation get agent in front)
- Comparison operators param server, macro actions
|
|
||
| def run(self, viewer = None, maintain_history = False, checkpoint_every=None): | ||
| results_tmp = ray.get([actor.run.remote(viewer, maintain_history, checkpoint_every) for actor in self.actors]) | ||
| def run(self, viewer = None, maintain_history = False, checkpoint_every=None, process_init_func=None): |
There was a problem hiding this comment.
what is process_init_func? is it a flag? or a function? can you add some documentation here? or add type hinting (see https://docs.python.org/3/library/typing.html)
| '//bark/runtime/tests:xodr_data', | ||
| ':params'], | ||
|
|
||
| deps = ["//bark/runtime/commons:parameters", |
There was a problem hiding this comment.
we should introduce some runtime lib definition (such as with geometry), so that we do not have to include them seperately every time.
| @@ -0,0 +1,141 @@ | |||
| # Copyright (c) 2020 fortiss GmbH | |||
There was a problem hiding this comment.
update copyright to 2021?
There was a problem hiding this comment.
(preferebly in a seperate commit)
| "PI means all angles considered", bark::geometry::B_PI*0.75); | ||
| max_lon_difference_to_be_front_ = params->GetReal( | ||
| "BehaviorIDMClassic::MaxLonDifferenceToBeFront", | ||
| "When are vehicles considered longitudinal-wise as front vehicles", |
There was a problem hiding this comment.
can you provide <> means ... as with the others?
| observed_world.GetAgentInFront(lane_corr); | ||
| observed_world.GetAgentInFront(lane_corr, GetMaxLatDifferenceToBeFront(), | ||
| GetMaxAngleDifferenceToBeFront(), | ||
| GetMaxLonDifferenceToBeFront() ); |
| @@ -40,6 +40,20 @@ BehaviorIDMStochastic::BehaviorIDMStochastic(const commons::ParamsPtr& params) | |||
| "BehaviorIDMStochastic::CoolnessFactorDistribution", | |||
| "From what distribution is the comfortable braking sampled in each " | |||
| "planning steo", | |||
| "UniformDistribution1D")), | |||
| use_intention_mechanism_(params->GetBool("BehaviorIDMStochastic::UseIntentionMechanism", | |||
| @@ -14,6 +14,12 @@ namespace bark { | |||
| namespace models { | |||
| namespace behavior { | |||
|
|
|||
| typedef enum YieldIntent { | |||
There was a problem hiding this comment.
consider using enum class, see https://stackoverflow.com/questions/18335861/why-is-enum-class-preferred-over-plain-enum
| EXPECT_TRUE(acc_primitive.IsPreConditionSatisfied(world1, corridors)); | ||
| auto traj3 = acc_primitive.Plan(0.5, world1, corridors.current); | ||
| EXPECT_NEAR( | ||
| traj3(traj3.rows() - 1, StateDefinition::X_POSITION), | ||
| traj3(0, StateDefinition::X_POSITION) + 5.0 * 0.5 + 4.0 / 2.0 * 0.5 * 0.5, | ||
| traj3(0, StateDefinition::X_POSITION) + 5.0 * 0.5 + 3.0 / 2.0 * 0.5 * 0.5, |
There was a problem hiding this comment.
make these numbers parameters so it is easier to understand what they mean?
| @@ -116,7 +116,7 @@ def test_const_label_function(self): | |||
| clf_after = pickle_unpickle(clf) | |||
| self.assertEqual(clf.value, clf_after.value) | |||
|
|
|||
| dlf = SafeDistanceLabelFunction("test", False, 1.0, 1.0, -5.0, -5.0, True, 4, False, 1.0, False) | |||
| dlf = SafeDistanceLabelFunction("test", False, 1.0, 1.0, -5.0, -5.0, True, 4, False, 1.0, 3.14, False) | |||
There was a problem hiding this comment.
use pi definition from bark geometry?
