diagnostics.recorder

Recorder: logs the run, plots it, and triggers the MuJoCo animation render.

Toggled by record:=true in the bringup launch. It subscribes to the joint states and the line-tension sensor, buffers the time series, and on shutdown:

1. Writes a PNG with joint positions, motor effort (torque) and the measured vs target line tension (so you can see if tension is held constant).

2. Saves the raw time series + the joint trajectory to an .npz file.

3. Auto-spawns the offscreen animation renderer (render_animation.py) under the .venv python, because MuJoCo lives there, not in the ROS python env.

Everything here uses only rclpy + numpy + matplotlib (all available to the ROS python), keeping the node light; the heavy MuJoCo rendering is a separate process fed the .npz.

Classes

Recorder

Functions

main([args])

Module Contents

class diagnostics.recorder.Recorder

Bases: rclpy.node.Node

save_and_render()

arm_joints

auto_render

eff

eff_t = []

js_t = []

mjcf_path

output_dir

qpos_joints

qpos_rows = []

render_script

ten = []

ten_t = []

ten_target = []

venv_python

video_format

video_fps

diagnostics.recorder.main(args=None)