apollo
apollo copied to clipboard
Is planning module deterministic?
Hi!
I want to ask about whether the output of the planning module is deterministic: given the same input, we always get the same output.
I found @jmtao 's comment at https://github.com/ApolloAuto/apollo/issues/11268#issuecomment-646842161
We use these test cases to make sure planning's output is fixed and determined based on the same set of input.
and I noticed some of those tests are actually disabled since 2019 https://github.com/ApolloAuto/apollo/commit/d524204e82927c384b555004f01c5293504d4de8 https://github.com/ApolloAuto/apollo/blob/986fe474272413fca0621b245d1c7aa6a1e75316/modules/planning/integration_tests/BUILD#L26
I ran the only enabled test sunnyvale_big_loop_test
for a couple times and had same results. I could not run other tests because I do not know how to solve those undefined references.
Questions:
- Is the output of planning module supposed to be deterministic?
- In the current version, is planning module deterministic?
- Will those disabled tests be fixed and enabled sometime soon?
Additional Question:
- How can I check coverage information of a test case? I tried
bazel coverage --instrument_test_targets --combined_report=lcov //modules/planning/integration_tests:sunnyvale_big_loop_test
and goterror: cannot open bazel-out/k8-fastbuild/bin/modules/planning/lattice/trajectory_generation/_objs/piecewise_braking_trajectory_generator/piecewise_braking_trajectory_generator.pic.gcno
I am not very familiar with Bazel and I am not sure about what I am missing.
- Some parameters are related with time, like stitching trajectory. Even if input message is same, but in different time planning start point is different.
- No. the reason is like before.
- We do not have this plan now. You can integrate test by Sim Control simulation.
@qwetqwe Hi! Thanks for responding. You can close this issue.