This thesis addresses accurate vehicle trajectory reconstruction at low speeds in GNSS-denied environments such as parking structures or indoor facilities. Pure IMU integration suffers from drift caused by bias accumulation and sensor noise. The research proposes combining IMU-based odometry with a kinematic steering model (e.g., bicycle model) to constrain vehicle motion and reduce estimation errors. The work will involve state estimation algorithms (e.g., Kalman filtering), bias modelling, and performance validation in low-dynamics scenarios. Drift reduction and accuracy improvements will be quantitatively assessed.
The thesis will be carried out in cooperation with a car manufacturer, providing experimental data and validation frameworks.
