Skip to content
Fun
research to playground model predictive control · steering dynamics · path tracking

MPC Path Tracking Lab

Two predictors steer identical vehicles with identical actuator dynamics. One assumes steering responds instantly; the other propagates a second-order steering model through its prediction horizon. Change speed and horizon length to compare path error and steering oscillation live.

IJAT 15 · 1155–1164 · 2014 Kim, E. · Kim, J. · Sunwoo, M.
Live comparison actuator-aware MPC vehicle-only MPC
vehicle speed
40 km/h
progress 0%
aware error 0.00 m
vehicle-only 0.00 m
Ready
Actuator-aware Case A
RMS error
0.00 m
Max error
0.00 m
Steer activity
0.0°/s
Vehicle-only Case B
RMS error
0.00 m
Max error
0.00 m
Steer activity
0.0°/s
Integrated prediction model

Both cars have identical steering actuator lag. The cyan predictor models that lag inside its horizon; the amber predictor assumes steering responds immediately.

angle constraint rate constraint receding horizon
δ command
→
2nd-order
steering
→
vehicle
motion
Tracking scope aware lateral error vehicle-only lateral error aware δ aware δcmd
Paper configuration

What the paper validated

The paper integrated an identified second-order steering model with a 2-DOF vehicle model, then formulated QP-based MPC with steering angle and rate constraints. DLC and road-course simulations used Np=20, Nc=8, Ts=10 ms, r=60 as the baseline configuration. The vehicle-only predictor developed increasing oscillation as speed rose.

Np
20
Nc
8
sampling
10 ms
steering BW
3 Hz
Simulation scope

A browser-scale conceptual model

This lab does not reproduce the paper's 27-DOF CarSim validation or Hildreth QP solver. It applies the same second-order steering lag to both cars and repeatedly evaluates finite-horizon candidates to visualize why actuator state belongs inside the predictor.

View the publication on About →