Real-Time Capable Simulation and Control of a Tractor-Implement System

Sammanfattning

This thesis presents a real-time capable simulation and control framework for a tractor-cultivator system using AGX Dynamics. The work includes modeling of a tractor, cultivator, tire dynamics, deformable terrain, and contact interactions. Different soil types and uneven terrain are considered to investigate how disturbances affect the system behavior. Two main control approaches are evaluated: a classical approach using Pure Pursuit and Proportional-Integral-Derivative (PID) control, and a model-based approach using a Linear Quadratic (LQ) controller. The controllers are evaluated with respect to cultivator path following, working depth control, and speed control when the tractor-cultivator system follows a predefined path. The results show that the LQ controller with integral action and feedforward achieves the best overall performance, with lower cultivator path following errors. The study also shows that AGX Dynamics is a suitable platform for real-time capable simulation of a tractor-cultivator system with deformable terrain interaction, while highlighting the trade-off between simulation quality and computational performance.