| Resumo: | This work presents a study on differential drive wheeled mobile robots regarding its localization estimation using sensor fusion techniques and control over a reference trajectory. The robot’s posture is an extremely important variable to be estimated, specially for autonomous mobile robots as it has to drive along a path without manual intervention. Posture estimation provided by individual sensors has shown to be inaccurate or straightforward mismatched, leading to a need of data fusion from different sources. Sensors such as accelerometer, gyroscope and magnetometer each have its own intrinsic constructive limitations. However, by combining all their flaws into a linear model allows optimal estimators, such as Kalman Filter (KF), to be used and produce estimates close to real life behavior. For the trajectory tracking and disturbance rejection, a linear control strategy for a linearized mobile robot model is applied. The system is modeled with error states to be carried out by a Linear Quadratic Regulator (LQR) controller along with a feedforward reference control action so that the reference trajectory is accordingly tracked.
|
|---|