Embedded Systems and Networking Approach to Guidance and Estimation
Abstract
Networked embedded systems are valuable tools for guidance and navigation, with applications ranging from UAV swarms to self-driving cars. LASR Lab is working on projects that require networked embedded systems running Kalman filters, such as a smart tower that will send an alert over the Internet if its embedded system detects structural changes in the tower. To demonstrate the architecture that will support these projects, a demonstration has been made consisting of a microcomputer and an Inertial Measurement Unit contained in a football-shaped shell. This football is designed to stream its position and orientation over a network in real time. Like other LASR Lab projects, the football utilizes a Tinker Board microcomputer and a VectorNav IMU.
Data are streamed in real time to a controller laptop over a Wi-Fi connection using the Open MPI protocol. An Extended Kalman Filter and an Unscented Kalman Filter for estimating position, velocity, orientation, and gyroscope biases were developed and implemented in C++, along with calibration routines for estimating initial conditions and noise parameters. Due to a lack of reliable measurements and mathematical bugs, the filters were not successful in estimating position, velocity, or attitude; but the hardware, software, and networking architecture was demonstrated successfully and can be used in future projects.
Citation
Ghan, Daniel Gonçalves (2019). Embedded Systems and Networking Approach to Guidance and Estimation. Master's thesis, Texas A&M University. Available electronically from https : / /hdl .handle .net /1969 .1 /186539.