Additional Sensors for Tello Drone

[Bubblun]

Hi Tello Pilots,

I am writing a master's thesis about the Tello drone, where we will try to make the drone fly autonomously in an indoor environment and execute different inspection tasks. I have some questions regarding this that I hope some of you have experienced before.

I plan to add a Raspberry Pi Zero to the drone to communicate with the drone and run the ROS NAV2 library. I may also add some additional sensors.

1. What sensors do you recommend adding to the Tello drone to make navigation and path planning easier?
2. What sensors are possible to add?
3. Have any of you tried mapping out the environment using different SLAM algorithms, what algorithm did you choose?

Best regards!

Bubblun

 

[Hacky]

Hi @Bubblun , welcome to the forum.
you will have to solve a lot of problems:

• Running ROS on the Pi Zero is almost impossible. There is an older port of ROS(1) that can be compiled to run under Raspian (but not the latest Raspi OS) and if you use it without graphical desktop, it fits inside the very limited RAM. Problem is, that most of the modules you plan to use (e.g. for SLAM) are very hard to compile and you'll find no precompiled binaries for the Raspi OS port. Many of them also require more RAM than you'll have on the Pi Zero
  The alternative is to use a "lightweight" Ubuntu based distribution like Ubuntu MATE. For this, almost any ROS version and module is available as binaries but you will not get this running on a Pi Zero.
• You will need at least a Pi Zero 2, as only this has 4 cores. Unfortunately, this also has only 512 MB RAM. Another Problem is, that it heats up very fast and requires additional cooling = additional weight
• The only realistic lightweight option to run ROS is a Raspi CM4. This has enough CPU power and enough RAM but requires additional cooling and an addition carrier board -> too heavy for the Tello.
• SLAM usually requires 360° lidar sensors. The lightest, that I found so far, weighs 49g -> too heavy

If you want to go a fully autonomous way, you should consider a larger frame. This is an example, where I made a retrofit based on an old AR.Drone 2.0 using ROS on a Raspi CM4 and a 360° lidar:

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

This is extremely compact and leightweight compared to other approaches I have seen so far but even this weighs already about 600g

In my opinion, you should try to use the Tello EDU as it is and try to live with the telemetry and camera streams, you can get via the SDK 3.0
The rest of the "intelligence" can run on a seperate system communicating with the Tello over WiFi. You see links to examples in my signature. There are also camera based SLAM algorithms but they require way more CPU/GPU power than you will ever be able to put on the Tello. You'll find some camera based SLAM examples with Tello on Youtube.

If you want to add sensors, in my opinion only a lightweight single beam LiDAR (like TF Mini or VL53L1X) makes sense in combination with a very leightweight ESP32 (like the Xiao ESP32C3) but this is not enough for SLAM.

 

[Bubblun]

@Hacky Why did you remove your reply? I found it very helpful!

 

[Hacky]

Why did you remove your reply? I found it very helpful!

I did'nt remove it. The forum software sent it into a moderator-approval-loop - don't know why but perhaps a moderator will explain it.

 

[Dan]

Hi Tello Pilots,

I am writing a master's thesis about the Tello drone, where we will try to make the drone fly autonomously in an indoor environment and execute different inspection tasks. I have some questions regarding this that I hope some of you have experienced before.

I plan to add a Raspberry Pi Zero to the drone to communicate with the drone and run the ROS NAV2 library. I may also add some additional sensors.

1. What sensors do you recommend adding to the Tello drone to make navigation and path planning easier?
2. What sensors are possible to add?
3. Have any of you tried mapping out the environment using different SLAM algorithms, what algorithm did you choose?

Best regards!

Bubblun

Hello. I like what you are doing. We now have a great need to fly drones in a cave on earth or the moon. You are ta higher level then we as we work mostly with high school students. We have worked with swarm flying, and various of hacking to include wireless charging charging, and RFID for logging flights and maintenance. We have hacked the ZUMI robot that uses a Raspberry Pi to maneuver, recognize QR codes, etc. We have also0 doe work with Aruco cards, and the DJI flight maps and cards, to navigate. We are presently working with moon maps to navigate these devices. We also have done some testing with teetered drones. You can see our work on Youtube MECATX, and contact me at [email protected]. We could do a ZOOM session. Much easier to communicate. Where are you located? We wauled like to follow your work. We have given up mostly on the Raspberry Pi ands switched to the Arduino and it's IDE in C++. We are learnings to write programs with Chatgpt and are getting ready fo the April total Eclipse. We are making Yagi antennas t o a ttached to a Celestron telescope mount and using Stellarium and CPWI to drive it with WiFi. We are using tow Dwarf2 telescopes to track the drones. At the next level we are using the Autel EVO 2 V3 with the live deck to transmit the video over Youtube live. I hope ths gives you some ideas.

 

[Hacky]

Meanwhile my initial answer became visible: Additional Sensors for Tello Drone

 

[sh3p]

Very cool! Interested to hear about how it is going as I am doing something very similar for my Master's thesis now! Was planning on testing the ORB_SLAM package with it in a few weeks. Have you tried this?

 

[Bubblun]

Very cool! Interested to hear about how it is going as I am doing something very similar for my Master's thesis now! Was planning on testing the ORB_SLAM package with it in a few weeks. Have you tried this?

Hi sh3p we are doing some initial testing with the orbslam3 now. Send me an email on [email protected] so we maybe can arrange a meeting