Robot terrestre para exteriores con navegación autónoma basado en fusión sensorial

Abstract

This research project focuses on developing an Ackerman kinematic ground robot for autonomous outdoor navigation based on sensor fusion navigation techniques, which are intended to reach a desired point. These techniques will focus on the use of stochastic estimators such as the extended Kalman filter (EKF) and trajectory tracking algorithms such as Pure Pursuit, which allow estimating the robot's position in real time and correcting its trajectory while heading to a desired point in such a way that the final position error is minimized. The sensors used for position and orientation estimation are: IMU (inertial motion unit), magnetometer and encoders, which provide relative motion-dependent measurements. On the other hand, the data from the sensors are obtained and processed through a small computer called Raspberry Pi 3 which in turn is implemented through the Python programming language algorithms for autonomous navigation and the generation of the actuator control signals is performed, for this type of robot the actuator used is a servo motor that controls the direction. All components are mounted on a chassis with Ackerman kinematics. The tests are performed on a level and obstacle-free terrain. Finally, the results of the position estimated by the Kalman Filter are compared with the position estimated by the Tracker software, thus obtaining a prototype capable of reaching a target zone of radius 60 cm in paths of up to 24 m, in addition to estimating trajectories with an RMSE error of up to 0.84 m.