| |
| Acadêmico(a): Fernanda Caroline Silveira Rodrigues |
| Título: Cálculo de odometria de um robô lego mindstorms através de rede neural |
| |
| Abstract: |
| One of the problems encountered in the context of autonomous robotic navigation its the ability of the system to estimate its own position. Thus, the robot has the possibility to check if his position is in the correct point of the traveled trajectory, allowing possible fixes. One of the forms for the robotic system to estimate his own position is when he leaves from of a known point and, from odometry values, estimates its new position. However, sometimes there are defects in the structure of the robot that are ignored by the odometry calculation, causing the unlimited accumulation of position estimation errors. This work presents the implementation of the odometry calculation and the Radial Function Neural Network Sensor Fusion method of as an alternative position estimation. The Radial Function Neural Network method combines the values of the odometer of the robot, based on LEGO Mindstorms platform, and a compass coupled to it, aiding the calculation of the estimated position. The advantage of using the Radial Function Neural Network in relation to odometry calculation is verified by comparing the distance of the estimated position and the actual position of the robot for each of the implemented methods. |
| |
|
