Arkaan QuanungaRoku · AI Innovation
Control & Systems EngineeringDec 2019Controls Engineer

AI Control System for an Autonomous Car

Sampled PID control with disturbance and noise models

Designed open-loop then closed-loop Simulink / MATLAB control systems for an autonomous-car simulation. Applied the Shannon–Nyquist theorem, modelled road curvature, wear, computational delay, and measurement noise, and settled on a sampled PID controller.

Sampled PIDController
±10%Fault Band
SimulinkTooling
The Problem & Engineering Constraint

The Core Challenge

A road simulation had to stay stable under wear and tear, uneven road, speed changes, computational delay, and ±10% coefficient faults — not just in the ideal open-loop case.
Technical Architecture & Approach

Engineering Solution & Implementation

Started with an open-loop Simulink model and MATLAB assist code, then closed the loop. Compared digital vs analogue control, added disturbance channels (road curvature, wear, delay), and established a sampled PID controller. Supervisor: Dr James E. Pickering.

Simulink — road curvature, wear, delay, and sampled plant with PID feedback
Simulink diagram of the autonomous-car control system
Road-profile / curvature against system output under disturbance
Plot of road curvature versus system output
Measured Production Impact

Verified Outcomes & Deliverables

Closed-loop sampled PID that corrects the car on the road simulation.

Modelled measurement noise and ±10% coefficient faults.

Documented Shannon–Nyquist sampling against total simulation time.

Technologies & Components

System Tooling & Technologies

MATLABSimulinkPID ControlOpen-loop SystemsClosed-loop Systems