Autonomous Drone - Wooded area navigation using GPS final location

[miniwheats231]

Background:

My team is building an autonomous drone that needs to be capable of navigating through several obstacles (like trees) to reach its final destination. They are meant to have swarm mentality, so ideally there will be several other drones making a mesh network with specific locations as well. For now, we are just making one autonomous drone that can make its way through the woods and get to its spot. The other "drones" will be sensors on sticks to show proof of concept. I personally don't have much experience in this, I am an undergrad in Mechanical Engineering, but I really want to learn. I have a pretty handy programmer on my team to assist. We only have three months to build the prototype so we are looking for feasible solutions for this time frame.


Questions:

We have been trying to get the flight controller to respond to the arduino using betaflight, but to no avail. We were able to get some response to the throttle, pitch, yaw, and roll when using the FrSky controller and a receiver. Does anyone have any suggestions to communicate to the flight controller? We plan on using Ultra sonic sensors for the tree avoidance and a barometer for the altitude.  

We are using an SBUS receiver with a Reaxacc F3+ OSD brushed flight controller (not using this for drone - just testing). Using an SBUS library by Brian Taylor. Both work fine when connected to the FrSky controller.

How do we communicate to the flight controller using an arduino?

[srnet]

Out of interest, when the GPS stops working because of the trees, how are you going to navigate ?

[miniwheats231]

As of now, we are assuming we wont loose GPS Signals... but if we do, we would use the localization of the other nodes to reference its required destination.

[ardly]

As of now, we are assuming we wont loose GPS Signals... but if we do, we would use the localization of the other nodes to reference its required destination.

So nodes 'A', 'B' and 'C' have GPS reception and know where they are. How does node 'D', without GPS reception, know where it is in relation to the other nodes and the destination?
Does the ultrasonics work for tree avoidance? I know bats can do it but I always thought they must be doing some pretty smart 3D processing since they can locate and catch insects.

[jremington]

we would use the localization of the other nodes to reference its required destination.

That sounds like an interesting challenge!

[mikb55]

If by drone you mean quadcopter then you are going to spend most of your time and all your money repairing it every time you hit a tree. It could be a very expensive project.
Suggest you start with a land vehicle. If you can dodge the trees then upgrade to something that flies.

[Bringamosa]

Avoiding obstacles with a drone is a big challenge.

If you have trouble getting just betaflight working, there will be many more obstacles to take witch much bigger challenges.

I have the feeling this is not really thought true.  How are you planning to avoid the trees?

[Bringamosa]

https://www.skydio.com/ have  a look at that

[miniwheats231]

So nodes 'A', 'B' and 'C' have GPS reception and know where they are. How does node 'D', without GPS reception, know where it is in relation to the other nodes and the destination?
Does the ultrasonics work for tree avoidance? I know bats can do it but I always thought they must be doing some pretty smart 3D processing since they can locate and catch insects.

Unfortunately I am not totally sure on the GPS question because that is a different teams part of the project. They are working on those kinds of problems while my team works on the hardware.

The Ultrasonics will be tested. Our adviser, who is a robotics engineer, seems confident they will work. So I am banking on that. But they will need to be tested. If they dont work we will come up with a different solution for obstacle avoidance.

Do you have any input on the question?

[miniwheats231]

If by drone you mean quadcopter then you are going to spend most of your time and all your money repairing it every time you hit a tree. It could be a very expensive project.
Suggest you start with a land vehicle. If you can dodge the trees then upgrade to something that flies.

Yes I am speaking of a quadcopter. We are having an impact halo around it to help reduce damage when it hits a tree. This project is a governmental one, which are typically capable of more expenses. They have reviewed our concept and have approved it knowing the risks.
We started with land vehicles but they cant navigate the terrain, which is why we chose flying.

Do you have any input on the question asked?

[Bringamosa]

You will get more response on the rcgroups forum I think from people with all the knowledge about quads.

[srnet]

Unfortunately I am not totally sure on the GPS question because that is a different teams part of the project. They are working on those kinds of problems while my team works on the hardware.

Whilst I understand that projects are split between various teams, there are certain issues that are fundamental to a project. In your case if the GPS does not work in your 'forest' then your project is a non-starter.
 
Given that its the work of a few minutes to connect a GPS to a Arduino, and check its working in the 'forest' has it been tested already ?

[jremington]

This project is a governmental one, which are typically capable of more expenses

Are we to believe that the "government" hires teams of complete amateurs, who don't talk to each other, and come to the Arduino hobby forum for advice?

[miniwheats231]

Are we to believe that the "government" hires teams of complete amateurs, who don't talk to each other, and come to the Arduino hobby forum for advice?

Believe it or not, they do assign projects to groups of students who are getting their degrees to learn something from them.Excuse me for trying to learn something new. So if you're not going to help, why even answer?

[jremington]

So learn from them, rather than post such astoundingly ignorant questions on a forum. You call into question the entire project.