Yi Han Poy: Sensor-Data Fusion of IMU and GPS Signal (Studienarbeit)

01.06.2021, 13:00 pm

Invitation to the intermediate presentation of Yi Han Poy

Topic: Sensor-Data Fusion of IMU and GPS Signal

Project: Research project

Supervisors: Sergio Petruz, Gökhan Akgün, Candy Lohse

Drones can perceive their surroundings with built-in sensors and thus fly autonomously. An important factor here is the localization of the drone. Measurement deviations can lead to inaccuracies. This causes the drone to navigate incorrectly and lead to a wrong destination. GPS sensors enable drones to determine their position accurately. However, they are subject to unavoidable errors when e.g. the satellites' reception is blocked. Using an additional sensor, i.e. the IMU sensor, can strengthen precision.

This project aims to implement a method to improve the position estimation capabilities of a drone. In this case, an IMU and GPS sensor should be used to determine accurately the position. The signals of the IMU sensor can normally be fused with Mahony and Madgwick filter, but not with the GPS signal. In the literature, an extended Kalman filter (EKF) is used to fuse the data of such sensors as it is also more accurate. Therefore, EKF should be implemented in this work.

This project will also exploit embedded architectures like PYNQ-Z1 in combination with ROS and simulation tools (Gazebo). This will allow a smooth transition of the system and field test. ROS allows integration into an environment that enables interfacing with other devices and algorithms such as the drone's flight controller. Therefore, the student should implement the sensor data fusion using IMU and GPS on the ARM Cortex A9 processor and validate the correctness of the implemented algorithms with dedicated test scenarios.