Working Notes · Robotics

Robotics Test Bench

Small simulation experiments for building physical intuition: make a prediction, run the smallest useful test, observe the result, then update the model.

  1. Question
  2. Prediction
  3. Experiment
  4. Observation
  5. Belief update

Resolved evidence

Experiments

#01Resolved

Feedback fundamentals

What does closing the loop change in a one-joint pendulum?

P-only feedback overshot or oscillated more; adding velocity feedback made the response settle more cleanly.

Inspect experiment →
#02Resolved

Multi-DOF dynamics: make coupling visible

What happens at one joint when only the other joint is driven?

A passive elbow moved with zero elbow command, replacing an independent-joint model with a full-mechanism view.

Inspect experiment →
#03Resolved

Model-based control: prepare torque from the model

What does dynamics knowledge buy, and what happens when it is wrong?

Accurate feedforward nearly removed the tracking residual. A controller-only mass error returned a structured feedback burden.

Inspect experiment →

View the canonical Robotics Test Bench repository ↗Code, issue state, and full reasoning history remain there.