2/6/25, 11:39 PM GitHub - saurabhmandlik/Vision-based-indoor-navigation-of-an-UAV

saurabhmandlik / Vision-based-indoor-navigation-of-an-UAV Public

0 stars 0 forks Branches Tags Activity

Star Notifications

Code Issues Pull requests Actions Projects Security Insights

3 Branches 0 Tags Go to file Go to file Code

saurabhmandlik Delete ss 64c37a1 · 4 years ago

Installation information Update README.md 5 years ago

New trained model Add files via upload 4 years ago

Project model Delete s 5 years ago

Scripts Delete code 4 years ago

Simple shape images Delete v 4 years ago

Thesis Work/Presentati… Delete ss 4 years ago

Tutorial of training mo… Create Objects1.py 4 years ago

images Create s 4 years ago

supervision added readme 4 years ago

README.md Update README.md 4 years ago

Referred links Update Referred links 4 years ago

Indoor non-GPS Autonomous Navigation of

UAV using AprilTag Marker
This repository explains the autonomous navigation and position of drone in indoor
environment. Here, I have used Intel RealSense camera to navigate the UAV using AprilTag
marker.

Required Set-up

https://github.com/saurabhmandlik/Vision-based-indoor-navigation-of-an-UAV 1/6
2/6/25, 11:39 PM GitHub - saurabhmandlik/Vision-based-indoor-navigation-of-an-UAV

Software and Dependencies to be needed

1. Ubuntu 18.04
2. ROS Melodic
3. Mission Planner
4. RViz
5. vision_to_mavros ROS packages and AprilTag marker
6. TensorFlow
7. Python 2.7

Hardware used

1. Raspberry Pi 4
2. Intel RealSense Camera
3. Raspberry Pi camera V2.0
4. UAV
5. PixHawk 4
6. RC transmitter
7. 4500 mAh LiPo Battery
8. StromPi and Battery module

The repository is divided into two parts:

1. Detect the predetermined object using computer vision approach

2. Autonomously navigate UAV using marker

2. Autonomously Navigate the UAV using AprilTag marker:

Few things we have to understand to get into this topic.

MAVROS refer this link

MAVROS is a ROS package that enables MAVLink extendable communication between

computers runnig ROS for any ground station. It's officially bridge between ROS and
ArduPilot by translating ROS topics into MAVLink messages.

Features of MAVROS

-Communication with autopilot via serial port, UDP or TCP (e.g. PX4 Pro or
ArduPilot)
-Plugin system for ROS-MAVLink translation
-Parameter manipulation tool
-Waypoint manipulation tool
-PX4Flow support (by mavros_extras)
-OFFBOARD mode support
-Geographic coordinates conversions.

MAVROS Installation

https://github.com/saurabhmandlik/Vision-based-indoor-navigation-of-an-UAV 2/6
2/6/25, 11:39 PM GitHub - saurabhmandlik/Vision-based-indoor-navigation-of-an-UAV

sudo apt-get install python-catkin-tools python-rosinstall-generator -y

mkdir -p ~/catkin_ws/src
cd ~/catkin_ws
catkin init
wstool init src
README

rosinstall_generator --rosdistro melodic mavlink | tee /tmp/mavros.rosinstall

rosinstall_generator --upstream mavros | tee -a /tmp/mavros.rosinstall

wstool merge -t src /tmp/mavros.rosinstall

wstool update -t src -j4
rosdep install --from-paths src --ignore-src -y

./src/mavros/mavros/scripts/install_geographiclib_datasets.sh

catkin build
source devel/setup.bash

Pixhawk FCU

Pixhawk is a flight controller unit that is used for many applications of multicopters, rovers
etc. It offers a wide variety of flight modes from fully manual to fully autonomous. As part of
the wider ArduPilot software platform it works seamlessly with a variety of Ground Control
Station programs that are used to setup the vehicle, monitor the vehicle’s flight in real-time
and perform powerful mission planning activities. For this project, I have used Pixhawk 4 FCU.

AprilTag Marker
Apriltag is
https://github.com/saurabhmandlik/Vision-based-indoor-navigation-of-an-UAV 3/6
2/6/25, 11:39 PM GitHub - saurabhmandlik/Vision-based-indoor-navigation-of-an-UAV

Overall Setup to Fly the UAV

Hardware Setup

-Connect Pixhawk FCU to LiPo battery.

-Connect FCU to on board computer raspberry pi through USB Port of FCU.
-Connect realsense camera to raspberry pi 4.
-Connect telemetry radio cable to PC. (Baud rate=57600)

Before doing this, check the following parameters in Mission Planner Software: (Information
regarding mission planner software and mandatory hardware configuration steps can
available in Thesis Work -> Mission Planner directory)

AHRS_EKF_TYPE = 2
BRD_RTC_TYPES = 2
EK2_ENABLE = 1
EK3_ENABLE = 0
EK2_GPS_TYPE = 3
EK2_POSNE_M_NSE = 0.1
EK2_VELD_M_NSE = 0.1
EK2_VELNE_M_NSE = 0.1
GPS_TYPE = 0
COMPASS_USE = 0
COMPASS_USE2 = 0
COMPASS_USE3 = 0
SERIAL5_BAUD = 921 (the serial port used to connect to Raspberry Pi)
SERIAL5_PROTOCOL = 1
SYSID_MYGCS = 1 (to accept control from mavros)
VISO_TYPE = 0

Ground test
Go to misssion planner software and click on the connect button (upper right part of the
window) with having baudrate 57600.

Steps to do on Linux Platform

We have to run three main nodes at the same time on 3 seperated terminal of RPi.

1. RealSense camera node:

- roslaunch rs_continous.launch

2. MAVROS node
- roslaunch px4.launch fcu_url:ttyACM0:57600

3. vision_to_mavros node:
- roslaunch vision_to_mavros tf_to_mavros_rs.launch

https://github.com/saurabhmandlik/Vision-based-indoor-navigation-of-an-UAV 4/6
2/6/25, 11:39 PM GitHub - saurabhmandlik/Vision-based-indoor-navigation-of-an-UAV

on another terminal of RPi, run command "mavros echo /mavros/vision_pose/pose" to see

the pose data from the vehicles (UAV).

and also run the "mavros echo /tf" to see the camera pose in Apriltag frame with Z-axis
pointing downwards. /tf is basically transforms the pose of source_frame_id and
target_frame_id which is defined in tf_to_mavros_rs.launch file.

The information regarding tf, vision_pose topic is available in "Thesis work->

vision_to_mavros" directory.

Check On Mission Planner Software

After running the all the nodes (realsense camera node, mavros node, vision_to_mavros
node), move to mission planner software and connect the battery.

I already mentioned above how to connect FCU to MP. After connecting to MP, You can see
the pop-up screen on MP window that shows getting parameters.

Then you have to set the EKF home position. It's shows the vehicle position in the world. For
that, I have a script set_origin.py where I have set SET_GPS_GLOBAL_origin and
SET_HOME_POSITION MAVLink messages. Run this scropt on terminal by command "rosrun
vision_to_mavros set_origin.py". Script available in visio_to_mavros -> scripts directory.

After running the script, you can see the quadcopter icon on MP software. (add SS here)

Now you are ready to fly UAV manually. Keep UAV on AprilTag marker and turn on the RC tx
ON and click on ARM button in MP. The rotors will start to rotate and according to given
thrust, we can manage the UAV using different modes. We will see the flight modes in details.

Releases

No releases published

Packages

No packages published

Contributors 2

saurabhmandlik Saurabh Mandlik

holger-ziti

https://github.com/saurabhmandlik/Vision-based-indoor-navigation-of-an-UAV 5/6
2/6/25, 11:39 PM GitHub - saurabhmandlik/Vision-based-indoor-navigation-of-an-UAV

Languages

Python 96.6% C++ 2.5% Other 0.9%

https://github.com/saurabhmandlik/Vision-based-indoor-navigation-of-an-UAV 6/6