Modeling and control for a lane keeping system in autonomous vehicles
This work presents the design, implementation, and experimental validation of multiple control strategies for an autonomous lane-keeping system based on a differential-drive mobile robot. Three controllers are investigated: a classical Proportional-Derivative (PD) controller, a State Space Feedback...
| Autor: | |
|---|---|
| Tipo de recurso: | tesis de maestría |
| Fecha de publicación: | 2025 |
| País: | España |
| Institución: | Universitat Politècnica de Catalunya (UPC) |
| Repositorio: | UPCommons. Portal del coneixement obert de la UPC |
| Idioma: | inglés |
| OAI Identifier: | oai:upcommons.upc.edu:2117/446276 |
| Acceso en línea: | https://hdl.handle.net/2117/446276 |
| Access Level: | acceso abierto |
| Palabra clave: | Automated vehicles Mobile robots Automatic control Vehicles autònoms Robots mòbils Control automàtic Àrees temàtiques de la UPC::Enginyeria electrònica |
| Sumario: | This work presents the design, implementation, and experimental validation of multiple control strategies for an autonomous lane-keeping system based on a differential-drive mobile robot. Three controllers are investigated: a classical Proportional-Derivative (PD) controller, a State Space Feedback (SSF) controller, and a Sliding Mode Controller(SMC).The objective is to evaluate their performance in trajectory tracking under real-world conditions. The PD controller provides a simple and intuitive solution, relying only on the tracking error and its derivative, and requiring minimal knowledge of the system dynamics. It is straightforward to tune and performs satisfactorily under nominaloperating conditions. However, its lack of robustness to modeling errors, sensor noise, and external disturbances limits its performance in more demanding scenarios. The SSF controller is a linear, model-based strategy that leverages a state-space representation of the robot’s kinematics. A Luenberger observer is integrated to estimate non-measurable states, such as the heading deviation angle, improving tracking precision. While this method allows full-state regulation in theory, its effectiveness strongly depends on the accuracy of the mathematical model. Experimentalresults revealed that modeluncertainties andobserver limitations significantly degraded performance, leading to its exclusion from the final comparative analysis. The SMC controller, based on variable structure control theory, is designed to enhance robustness against modeling errors, friction, and external perturbations. By driving the system dynamics toward a predefined sliding surface and maintaining motion along it, SMC achieves asymptotic stability and accurate trajectory tracking despite parameter variations and disturbances. A saturation function was introduced to replace the traditional sign function, effectively mitigating chattering effects commonly associated with SMC. All controllers were initially validated in simulation using Matlab/Simulink before real-world implementation on a Texas Instruments TMS320F28379D microcontroller. Experimental results show that both the PD and SMC controllers successfully follow the desired trajectories on two different test circuits: an oval path and a sinusoidal path. The performance of the controllers was quantitatively assessed using the Integral of Time-weighted Absolute Error (ITAE) of the lateral deviation with respect to the reference path. Results demonstrate that the PD controller achieves lower lateral deviation errors compared to SMC. The performance gap is attributed to suboptimal tuning of the SMC parameters and a potentially inadequate selection of the sliding surface σ, which may have limited its ability to minimize tracking errors effectively. |
|---|