Transforming DJI Tello into an Autonomous AI Tracking Drone

Table of Contents

1. Introduction
2. Components for the Project
3. DJI Tello: A Small Quadcopter with Impressive Features
4. Tello SDK: An Overview of DJI's Software Development Kit
5. YOLO Darknet: A Powerful Object Detection Framework
6. Setting Up the Tello Command Script
7. Receiving Sensor State from Tello with Tello State Script
8. Streaming Video from Tello with Tello Stream Script
9. Training YOLO Darknet for Object Detection with Tello AI
10. Controlling Tello's Movement with Machine Learning
11. Conclusion

Introduction

Welcome to this article where we'll explore how to transform a DJI Tello quadcopter into a drone that can follow you anywhere. We'll cover the necessary components for this project, dive into the features of the DJI Tello quadcopter, explore the Tello SDK for communication and control, discuss the powerful YOLO Darknet object detection framework, and provide step-by-step instructions on setting up scripts to command the Tello, receive sensor state, stream video, and create an AI model for object detection. So let's get started on this exciting journey of turning a quadcopter into an autonomous drone!

Components for the Project

To begin our project of transforming the DJI Tello quadcopter into a drone that follows you, we need to understand the key components involved. These components include the machine learning framework, DJI Tello quadcopter, sensor module, command module, video module, logging system, control system, and state machine inside a Python backend. Combining these components will allow us to assemble and control the quadcopter effectively. Let's delve deeper into each component to understand its role in this exciting project.

Machine Learning Framework

The machine learning framework will serve as the backbone of our project. It will be responsible for training and deploying the object detection model that the DJI Tello quadcopter will utilize to track and follow objects or individuals. We'll be using the powerful YOLO Darknet framework for this purpose.

DJI Tello Quadcopter

The DJI Tello quadcopter is a compact and lightweight drone that provides an ideal platform for our project. With its reasonable price tag of $99, it offers impressive features such as a 13-minute flight time, 100-meter range, 720p camera, and collision detection system. The Tello also incorporates a vision positioning system and offers SDK support for customization and control.

Sensor Module

The sensor module plays a crucial role in providing real-time data about the quadcopter's environment. It includes sensors such as a barometer and pressure sensor for altitude measurement, a Time-of-Flight (ToF) sensor for depth sensing, and battery percentage and Wi-Fi signal strength indicators.

Command Module

The command module enables communication with the DJI Tello quadcopter. It establishes a connection using UDP (User Datagram Protocol) and allows us to send commands to the Tello and receive responses. This module is essential for controlling the quadcopter during autonomous flight.

Video Module

The video module enables streaming video from the DJI Tello quadcopter. By accessing the video feed, we can monitor the drone's perspective and utilize it for object detection and tracking.

Logging System

The logging system enables the collection and storage of data generated during the quadcopter's flight. It provides valuable insights into various aspects of the drone's behavior and aids in analyzing and optimizing its performance.

Control System

The control system encompasses the algorithms and logic responsible for controlling the DJI Tello quadcopter's movements. It utilizes the sensor data, commands, and machine learning model to make informed decisions regarding the drone's flight path and actions.

State Machine Inside Python Backend

The state machine, implemented within the Python backend, keeps track of the quadcopter's current state and manages the flow of commands and responses. It ensures smooth coordination between the different components and facilitates efficient communication and control.

Now that we have a clear understanding of the components involved in this project, let's explore the DJI Tello quadcopter and its features in more detail.

DJI Tello: A Small Quadcopter with Impressive Features

The DJI Tello quadcopter, developed in collaboration with Ryze Robotics, offers remarkable features considering its compact size. With a weight of just 80 grams, the Tello is small but packed with impressive capabilities. Let's take a closer look at the specifications and features of the DJI Tello quadcopter.

• Compact Design: The Tello is a small quadcopter that is both lightweight and portable. Its diminutive size makes it easy to carry and maneuver.
• Intel Processor: Powered by an Intel 14-core processor, the Tello offers robust computational capabilities necessary for stable flight and intelligent features.
• Collision Detection System: The Tello quadcopter is equipped with a collision detection system. This innovative technology allows the drone to detect potential collisions and automatically shut off the motors to prevent damage.
• Vision Positioning System: Utilizing DJI's advanced vision positioning technology, the Tello can accurately determine its position and maintain stability even without GPS signals.
• High-Quality Camera: The Tello features a 720p camera, offering a good balance between image quality and real-time video streaming performance. It allows you to capture detailed aerial footage during the drone's flight.
• Extended Flight Time: With a flight time of up to 13 minutes, the Tello provides ample opportunity to explore and Record your surroundings from an aerial perspective.
• Impressive Range: The Tello has a range of up to 100 meters, allowing for extended flights and broader exploration.
• SDK Support: The Tello offers an extensive software development kit (SDK) that provides developers with access to various functions and capabilities. This SDK enables customizations and the integration of additional features into the drone.

At an affordable price of $99, the DJI Tello quadcopter offers a compelling proposition for enthusiasts and developers alike. Its small form factor, robust features, and SDK support make it an ideal choice for projects like turning it into a drone that can autonomously follow you. Let's delve into the Tello SDK and explore its capabilities.

Tello SDK: An Overview of DJI's Software Development Kit

The Tello SDK (Software Development Kit) provides developers with a comprehensive set of tools and resources to customize and interact with the DJI Tello quadcopter. By leveraging the Tello SDK, developers can command the quadcopter, receive sensor data, capture video streams, and even create their own applications for controlling the drone. Let's dive into the details of the Tello SDK and understand how it facilitates communication and control.

The Tello SDK allows communication with the DJI Tello quadcopter using UDP (User Datagram Protocol). It offers a straightforward and versatile method for sending commands and receiving responses. To establish communication, you need to specify the quadcopter's IP address, which is typically 192.168.10.1, and the respective port numbers for sending commands (8889) and receiving responses (9000).

Using the Tello SDK, you can send various commands to the quadcopter, such as takeoff, land, move left, move right, move forward, move backward, and so on. The SDK provides a plain and simple interface, making it easy to control the Tello's movements and actions.

Additionally, the Tello SDK offers access to a Tello state, which allows you to stream sensor data from the quadcopter. The sensor data includes information such as barometer readings, pressure sensor data, battery percentage, Wi-Fi strength, and more. By tapping into the Tello state, you can Gather real-time information about the quadcopter's internal state and make informed decisions based on the sensor readings.

The Tello SDK also enables video streaming from the Tello quadcopter. By utilizing the respective IP address and port number, you can seamlessly receive video frames from the Tello and display them using frameworks like OpenCV. This capability opens up possibilities for real-time video analysis, object detection, and more.

Overall, the Tello SDK empowers developers to interact with the DJI Tello quadcopter in a seamless and customizable manner. Its UDP-based communication, command structure, access to sensor data, and video streaming capabilities provide a robust foundation for creating innovative applications and projects. Now that we have a good understanding of the Tello SDK, let's explore the YOLO Darknet framework, which will be instrumental in our project.

YOLO Darknet: A Powerful Object Detection Framework

YOLO (You Only Look Once) Darknet is a state-of-the-art object detection framework known for its speed and accuracy. Developed by Joseph Redmon and further improved by AlexeyAB, YOLO Darknet can detect multiple objects in an image with impressive real-time performance. Let's delve into the details of YOLO Darknet and understand why it's the perfect choice for our project.

Key Features of YOLO Darknet

• Fast Object Detection: YOLO Darknet offers real-time object detection, even on modest hardware configurations. It can process frames at high speed, making it ideal for applications that require quick and precise object recognition.
• Detection of Multiple Objects: One of the unique selling points of YOLO Darknet is its ability to detect multiple objects simultaneously. It can identify up to 9000 objects in a single image, making it suitable for situations where object density is high.
• mAP (mean average precision) and FPS (frames per Second): YOLO Darknet achieves a good balance between object detection accuracy and speed. It achieves competitive mean average precision scores while maintaining high frames per second rates. This ensures accurate detection without significant performance trade-offs.
• YOLOv3 Tiny: For our project, we'll focus on YOLOv3 Tiny, a compact version of the YOLO model. While it may have slightly lower mean average precision, it excels in terms of frames per second, making it ideal for low-end GPUs or systems with limited computational resources.

Now that we have an overview of YOLO Darknet and its capabilities, let's proceed to set up and command the DJI Tello quadcopter using Python scripts.

Setting Up the Tello Command Script

To communicate with the DJI Tello quadcopter and send commands via UDP, we'll create a Python script called "tello-command.py." This script will serve as the bridge between our commands and the quadcopter's actions. Let's dive into the details of setting up this script.

import socket

# Create a UDP socket
sock = socket.socket(socket.AF_INET, socket.SOCK_DGRAM)

# Set the Tello address
tello_address = ('192.168.10.1', 8889)

# Bind the socket to a local address, so we can receive the response from the Tello
sock.bind(('', 9000))

# Send a command to the Tello
message = 'command'
sock.sendto(message.encode(), tello_address)

# Listen for responses
while True:
    try:
        # Receive the response from the Tello
        response, _ = sock.recvfrom(1024)
        print(f'Response from Tello: {response.decode()}')

        # Check if the response indicates the end of the communication
        if 'end' in response.decode():
            break

    except KeyboardInterrupt:
        # Close the socket if any exception occurs or the user interrupts the program
        sock.close()
        break

The above code snippet shows the basic structure of the "tello-command.py" script. Here's what the script does:

1. It imports the necessary socket module to establish a UDP connection.
2. It creates a UDP socket using the socket.AF_INET and socket.SOCK_DGRAM parameters.
3. It sets the Tello's IP address (usually 192.168.10.1) and the respective port number (8889) for sending commands.
4. It binds the socket to a local address using the bind() method, allowing us to receive responses from the Tello.
5. It sends a command (in this case, 'command') to the Tello using the sendto() method. We encode the message as bytes before sending it.
6. It enters a while loop to listen for responses from the Tello.
7. It receives the response from the Tello using the recvfrom() method and prints it to the console.
8. It checks if the response contains the keyword 'end' to determine the end of communication.
9. If an exception occurs (e.g., KeyboardInterrupt) or the user interrupts the program, it closes the socket and exits the loop.

With this script in place, you can easily send commands to the DJI Tello quadcopter and receive responses. Feel free to modify the commands and explore the capabilities of the Tello. In the next section, we'll create another Python script to receive sensor state from the quadcopter.