Lab 2 — Example Results

Week 7 · Spring 2026

Example step-response figures from the three experiments (Base, FeedForward, Tuned) for the speed and yaw-rate channels. Each case has two figures:

Timeseries — setpoint, measured response, and actuator command across the full step window.
PID internals — Tar / Act with metrics text box (top), and the per-term breakdown (P, I, D, FF) plus the actuator command (bottom).

A separate integrator-windup demonstration on the Base/yaw run shows what happens when the controller is allowed to hold the step long enough for the I term to wind up.

Speed Channel

Base - Speed

ArduRover stock PID gains, FF off, rate limiters off, throttle baseline off. The reference against which FeedForward and Tuned are compared.

Base / speed — PID internals (Tar/Act with metrics; P/I/D/FF terms with actuator command)

Feedforward - Speed

Tuned - Speed

Yaw-rate (Steering) Channel

Base - Yaw-rate

Known setup issue in this dataset: the baseline parameter file shipped with ACRO_TURN_RATE = 360 deg/s (≈ 6.28 rad/s), well above what the boat can physically achieve. Full-stick yaw commands therefore mapped to an unrealistically large setpoint, and the measured yaw rate plateaus far below the commanded value — visible as a large steady-state error in the figures and the metrics table. The procedure now has students reduce ACRO_TURN_RATE to a value the boat can actually track.

Base / yaw — timeseries (setpoint, measured EKF yaw rate, rudder command)

Base / yaw — PID internals (Tar/Act with metrics; P/I/D/FF terms with rudder command)

Base - Yaw-rate (integrator windup demonstration)

The same Base/yaw configuration as above, but the step input is held long enough for the integrator to wind up. The delayed recovery on the step-back illustrates classic I-term windup.

Base / yaw / windup — timeseries with extended step hold

Base / yaw / windup — PID internals; note the I term grows for the duration of the step and dominates the recovery transient

Feedforward - Yaw-rate

Tuned - Yaw-rate

Summary

Tuned PID gains and step-response metrics across the three experiments for both channels. Generated by closedloop_assess_main.m; the table content is included from metrics_summary.md, regenerated each time the analysis script runs.

label	P	I	FF	RiseTime (s)	SettlingTime (s)	Overshoot (%)	SS Error
base_speed	0.2	0.2	0	2.030	20.293	11.50	0.0000
ff_speed	0.2	0.2	0.2	2.791	10.079	0.53	-0.0200
tuned_speed	0.4	0.1	0.2	5.521	10.139	0.44	0.1600
base_yaw	0.2	0.2	0	2.608	NaN	19.88	1.3014
ff_yaw	0.4	0.1	1.2	1.121	NaN	27.43	-1.4310
tuned_yaw	0.4	0.1	1.2	1.181	NaN	30.04	-1.3611