Vehicle Controller Frequency Response & Pseudo-Chirp Analysis

Domain

Autonomous Vehicle Dynamics & Control, System Identification

Tools

Python (SciPy/NumPy), MATLAB/Simulink, Signal Processing, Cubic Splines

The engineering challenge

Evaluating the dynamic response and bandwidth limits of a vehicle steering or velocity controller requires testing across a wide frequency range. Standard linear sine sweeps present two problems: large amplitudes at low frequencies cause excessive vehicle displacement, quickly running out of track, while abrupt frequency steps create acceleration and jerk spikes that trip actuator torque limits and safety-controller overrides.

Engineering decisions & implementation

To excite the closed loop safely across the full target bandwidth, I built a custom excitation and analysis pipeline:

[ Raw Frequency Target ]
        │
        ▼
[ C² Spline Smoothing ]
        │
        ▼
[ A/ω² Amplitude Scaling ]
        │
        ▼
[ Safe Vehicle Input ]

• C² continuity via cubic splines — smooth position, velocity, and acceleration across input transitions to eliminate artificial jerk spikes that trip steering or drive actuators.
• Dimensionless A/ω² displacement scaling — command amplitudes scaled down at higher frequencies to bound physical path displacement while keeping enough excitation to analyse tracking performance.
• Automated Bode & phase-lag extraction — a script that processes commanded-vs-measured telemetry to extract gain margin, phase lag, and closed-loop bandwidth.

Results

• Hardware-safe sweeps — dynamic frequency sweeps run on vehicle control loops without triggering over-torque faults or exceeding track lateral boundaries.
• Automated diagnostics — replaced manual, single-frequency test runs with a pipeline that outputs the controller's empirical Bode plot and tracking limits directly from test logs, capturing the −3 dB bandwidth and phase margin.