TUM-Aries-Lab / hip-controller / 34614829761
86%

DEFAULT BRANCH: main
Ran
Jobs 1
Files 30
Run time 1min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst
coverage: 83.815% (+0.08%) from 83.739%
34614829761

push

github

NGierden 
Add property tests for the sensor preprocessing pipeline

SensorPreprocessor had no direct test. It sat at 78% coverage purely
incidentally, via the end-to-end controller tests, and those cannot see
whether the stages are actually wired together.

That gap is not hypothetical. An earlier refactor of this class ended up
constructing two SogiFllFiltering objects: one filtering the angle, and a
second that set_locomotion_mode / set_demo_mode configured but which was
never stepped. Per-mode retuning silently became a no-op, and on the SOGI
velocity path the output velocity was identically zero. The entire suite
still passed, because the end-to-end tests only assert the command is finite
and inside the motor limits -- which zero velocity does not violate.

Verified these tests close that hole by mutating the source and re-running:

  velocity forced to zero    -> 3 failures, including both velocity tests
  mode switches made no-ops  -> 2 failures, both mode-change tests

Covered: velocity is non-static and in phase with the true derivative, the
filtered angle tracks the input, a constant mounting offset is removed,
set_locomotion_mode and set_demo_mode both change the filtered output, the
stand-still gate decays the oscillator, reset restores initial response,
extreme inputs stay finite, non-monotonic timestamps raise, and the first
sample passes through unchanged.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

1103 of 1316 relevant lines covered (83.81%)

0.84 hits per line

Jobs
ID Job ID Ran Files Coverage
1 34614829761.1 30
83.81
GitHub Action Run
Source Files on build 34614829761