Presentation Title

Titan Providence

Faculty Mentor

Yu Bai

Start Date

23-11-2019 8:45 AM

End Date

23-11-2019 9:30 AM

Location

174

Session

poster 2

Type of Presentation

Poster

Subject Area

engineering_computer_science

Abstract

Project Title: Titan Providence: Self-navigation drone follows the way the of GPS

Student Names: Emilio Correa, Tanis Yusico-Suarez, Marcos Reyes

Affiliated Department: Computer Engineering Program

Project Advisor: Dr. Yu Bai

Data and analysis of information have become a crucial aspect in societies current development, allowing accurate awareness of our environments, benefitting maintenance, safety, and overall innovation. However, being informed in some situations require extensive and intensive manpower and are generally costly and dangerous, with situations ranging from inspecting power lines to plotting potholes within an urban area’s road system. Titan Providence utilizes a drone and it’s improved accessibility and maneuverability to enable applications in various fields that would otherwise be prohibitive given previous limitations. Differing from common Unmanned Aerial Vehicle (UAV) systems, Titan Providence implements localized machine learning, such as GPS, allowing for fully autonomous completion of extensive analysis of GPS coordinates which allow for the drone to enter, vast, hazardous, monotonous environments, or a combination thereof, in addition to having the benefits of an overall lowered cost in finance and labor. Our objectives for this project including construction and design of the drone framework, adaptive pathfinding, and awareness, in addition to task assignment.

To achieve our goals, we integrated a Raspberry Pi 4 with a compatible flight control module (Pixhawk PX4) with additional telemetry and peripherals docked on a custom quadcopter carbon fiber chassis that allows modularity and expansions in accordance to achieve such task as moving the drone from one location to another, with the use of coordinate as the location variables. By using these two systems in conjunction, we passively maintain the stability of the system through its inertial measurement unit (IMU) and it’s 4 individual multirotor electronic speed controllers (ESC), allowing further independence of the system’s microcomputer. In regards to the use of GPS navigation, we have tried to implement Deep Neural Network (DNN), in which the system reads the GPS location, taken from the Pixhawk module, calculates the angle of the two coordinates, having the drone match the set parameters for which the drone must follow. From this, we were able to have Titan Providence confidently identify paths to follow on an unknown environment with GPS data through its ability to decide which way the path was leading. Additionally, our system architecture allows us to further script flight controls using Python, providing general mission assignments to the drone outlined by the user that is fully localized and that can be initialized through SSH using a cellphone or computer. To begin testing, we have implemented several programs into the Raspberry Pi 4, converting the GPS information received by the Pixhawk, such information as longitude, latitude, and altitude. This information is then set as variables, implemented into the program which would have the drone move in a certain way based on restrictions given already to the pixhawk. We have had several failures and successes, one of those successes leading to us understanding much deeper the concept of the motors, using such functions as set_attitude, which has helped control the motors in keeping the same altitude, moving forward and backward etc. The latest development of our research is constructing a code to turn the drone to face the designated location, through comparison of angles of the front of the drone and angle of the drone’s location to designated one. We now seek to improve the maneuvering of the drone in order to implement image based sensors to determine the most effective path.

This document is currently not available here.

Share

COinS
 
Nov 23rd, 8:45 AM Nov 23rd, 9:30 AM

Titan Providence

174

Project Title: Titan Providence: Self-navigation drone follows the way the of GPS

Student Names: Emilio Correa, Tanis Yusico-Suarez, Marcos Reyes

Affiliated Department: Computer Engineering Program

Project Advisor: Dr. Yu Bai

Data and analysis of information have become a crucial aspect in societies current development, allowing accurate awareness of our environments, benefitting maintenance, safety, and overall innovation. However, being informed in some situations require extensive and intensive manpower and are generally costly and dangerous, with situations ranging from inspecting power lines to plotting potholes within an urban area’s road system. Titan Providence utilizes a drone and it’s improved accessibility and maneuverability to enable applications in various fields that would otherwise be prohibitive given previous limitations. Differing from common Unmanned Aerial Vehicle (UAV) systems, Titan Providence implements localized machine learning, such as GPS, allowing for fully autonomous completion of extensive analysis of GPS coordinates which allow for the drone to enter, vast, hazardous, monotonous environments, or a combination thereof, in addition to having the benefits of an overall lowered cost in finance and labor. Our objectives for this project including construction and design of the drone framework, adaptive pathfinding, and awareness, in addition to task assignment.

To achieve our goals, we integrated a Raspberry Pi 4 with a compatible flight control module (Pixhawk PX4) with additional telemetry and peripherals docked on a custom quadcopter carbon fiber chassis that allows modularity and expansions in accordance to achieve such task as moving the drone from one location to another, with the use of coordinate as the location variables. By using these two systems in conjunction, we passively maintain the stability of the system through its inertial measurement unit (IMU) and it’s 4 individual multirotor electronic speed controllers (ESC), allowing further independence of the system’s microcomputer. In regards to the use of GPS navigation, we have tried to implement Deep Neural Network (DNN), in which the system reads the GPS location, taken from the Pixhawk module, calculates the angle of the two coordinates, having the drone match the set parameters for which the drone must follow. From this, we were able to have Titan Providence confidently identify paths to follow on an unknown environment with GPS data through its ability to decide which way the path was leading. Additionally, our system architecture allows us to further script flight controls using Python, providing general mission assignments to the drone outlined by the user that is fully localized and that can be initialized through SSH using a cellphone or computer. To begin testing, we have implemented several programs into the Raspberry Pi 4, converting the GPS information received by the Pixhawk, such information as longitude, latitude, and altitude. This information is then set as variables, implemented into the program which would have the drone move in a certain way based on restrictions given already to the pixhawk. We have had several failures and successes, one of those successes leading to us understanding much deeper the concept of the motors, using such functions as set_attitude, which has helped control the motors in keeping the same altitude, moving forward and backward etc. The latest development of our research is constructing a code to turn the drone to face the designated location, through comparison of angles of the front of the drone and angle of the drone’s location to designated one. We now seek to improve the maneuvering of the drone in order to implement image based sensors to determine the most effective path.