Netherlands
Asked — Edited

Autonomous Parrot Ar 2 Drone Navigation

Hey everybody!

First of all, I am amazed by how good your software works! I've been searching for software to use in a project for my university and ARC is by far the most promising software i've found.

So I have a few questions that i'd hoped you guys could help me with. For a project on my university me and a few fellow students want to use a drone to help people navigate within the school. To do this we are using a Parrot AR 2 drone, together with (hopefully) ARC. We have played with ARC for a bit to try and understand how the software really works. This went well until the point that we have to make the drone fly autonomously, why autonomously you might think? We want the drone to be able to fly to a given location within the school, by itself, without any user interaction whatshowever AND without crashing or being a hazard to people. This is where we run into trouble. We got the drone to perform a script, but for it to fly autonomously we need the drone to recognize 'danger' being people (automatic ascending) and close proximity to walls (automatic centering in hallways). Do you guys think this is possible to do with ARC, and if so, can you give us a push in the right direction?

Besides this; is it possible to combine different functionality of ARC? For example, let the drone perform a script, while using the camera to make sure it doesn't hit anything, give a led signal when the destination has been reached and then fly back to the starting point?

Im very curious what you guys think about this and if somebody is willing to help us out a bit :D

Lets end this post with another (small) question!: After I connect to the drone in ARC, and let it take off using my keyboard and then land again, the drone won't respond to any scripts I send to it, it will only listen to my keyboard and I have to reset the connection for it to execute scripts again, is there a way to fix this, so to make the drone respond to everything I tell it to do?

Sorry for the long post!


ARC Pro

Upgrade to ARC Pro

Experience early access to the latest features and updates. You'll have everything that is needed to unleash your robot's potential.

#1  

Autonomous indoor navigation with the parrot drone.... Hmmmm as good as ARC is, currently (unless you are a engineer at google) I am going to go out on a limb and say this is not possible... With GPS it is, but GPS is only accurate within 10ft.... so that would not be good enough for indoor navigation.... If you want to navigate indoors, you are going to need a human pilot....

PRO
Synthiam
#2  

The challenge is using a camera to detect distance - is kind of impossible... at least without knowing how to identify an object to determine how close you are too it.

I have a GPS Flight Recorder for the AR Drone v2 - it weighs about as much as the EZ-B v4 without the plastic shell. You can use the EZ-B v4 without a plastic shell and connect Ultrasonic Sensors to the drone's body.

There is already a control in ARC that allows that:)

User-inserted image

Netherlands
#3  

@Richard How about the use of some sort of Wi-Fi positioning system, would that be good enough for a parrot drone to determine it's location in your opinion?

@DJ Sures The thought of additional proximity sensors places on the indoor hull crossed my mind before, though I have no clue on how to connect these to the drone for it to be able to use data from them. Also, due to safety, we are required to keep the indoor hull on at all times when flying with the drone, or do you mean to keep the hull on and also place the EZ-B v4 on the drone? Won't that make the drone unstable, especially when it has to fly through hallways that are 2 metres wide? It's a good suggestion, thanks! But, is it safe to say its near impossible to prevent collisions with walls using sensors already in the drone or ARC?

#4  

What wifi positioning system? I have never heard of anything like that... Do you have a link? You are probably new to robotics because otherwise you would know this an unbelievably difficult to accomplish task... The equivalent question would be... "can I build a car with a 3d printer?" The answer... is yes it is possible, but what a monumental task it would be....

In order for the drone to navigate indoors, it needs to know where it is in it's surroundings... This is too big, but it might work... Laser range finder Neato floor vacuums use this technology, however they only need to know x and y coordinates... The drone needs to know x, y and z....

PRO
Synthiam
#5  

@yuckoz the EZ-B v4 without the plastic shell weighs about the same or less than the GPS Flight Recorder. This allows you to mount the EZ-B v4 on the AR Drone. You may also connect Ultrasonic Distance sensors to the AR Drone for distance avoidance. The sensors are connected to the EZ-B v4. The drone on it's own is unable to avoid obstacles or detect distances.

Netherlands
#6  

@Richard R I meant something like this, though I'm not sure if it's possible for the drone to work with a system of that kind.

@DJ Sures Thanks for the good idea's. I will discuss it with my teammates, problem is that we don't really have a budget, so it will be tricky. Seems that we will need to have some sort of additional piece of hardware for our plan to be able to succeed, the drone can do much, but not all.

#7  

It doesn't say much, but I got the gist of it... It didn't mention accuracy, but for indoor use I imagine you will need accuracy within at least a few inches...

I know EZ Robot (DJ) is rumoured to be working on a camera based indoor navigation system... I don't know any more details than that and I don't know if it will be able to work with flying type robots, but it sounds promising... I guess we'll have to stay tuned...:)

Netherlands
#8  

I've been looking at some of the script functions within ARC and my eyes got caught by the label and Goto (label) parts. Is it possible to define labels in a script, and make the drone fly from label to label to get to the destination of the user, where the user can input his destination and the script 'calculates' which route the drone should take?

Ergo: Is it possible for me to have for example, 30 labels, and write a short script to every possible destination?

Example: User wants to go to destination 1: TakeOff(); Sleep (3000); Goto (label 1); <- Starting point Goto (label 3); Goto (label 4); Sleep (5000); <- waiting at the destination for a bit so the user knows Goto (label 3); <- starting journey back to starting point Goto (label 1); <- back at starting point Land();

Some same script to go to destination 2, 3, 4 and so on flying from label to label and then returning by flying from label to label.

I'm asking because we need to start somewhere, and build up from that point. So if we at the very least can get the drone to fly regular patterns, we can step by step improve the quality of the flight by maybe adding sensors or other possible solutions.