[Discourse.ros.org] [ROS Projects] Quadcopter shooting high power lasers controlled with motion capture (Otus tracker)
Hi! This project will document my attempts at shooting at baloons with a high power laser from an autonomous quadcopter. I will provide the documentation to replicate the whole system. I already made a simple quadcopter controller that can follow a path as you can see there:
I am the lead developer of the motion capture system [Otus tracker](https://www.rcbenchmark.com/otus-tracker/), and the mechanical developer of the dynamometer for motors and propellers RCbenchmark [Series 1580](https://www.rcbenchmark.com/dynamometer-series-1580/). A collegue made a ROS plugin for the Otus tracker and software. I learned ROS to code the quadcopter controller. We also coded another example with a ground small robot following another tracker with ROS:
First, I want so talk about safety. High power lasers are dangerous and can cause permanent eye damage, fire and burns. We are performing all tests in a closed, locked room with safety glasses. If you work with lasers, make sure you have quality safety glasses. For this project, I will be using a high power laser and an off the shelf quadcopter.
Currently, I am sending the command through an arduino acting as a ROS node. The Arduino then sends the command to a receiver, and finally to the quadcopter. The advantage of this method is that an operator can still control the UAV by simple releasing a switch from the controller. The disadvantage is that the update rate is limited to 50 Hz, the setup adds latency. Consequently, the quadcopter cannot be controlled as precisely. The Otus has a resolution of 1 mm or better and a latency under 3 ms, but the latency from the uplink prevents us from using a controller that is too aggressive. If this causes problem, I will switch to Xbee communication.
I will be using a 500 mW laser to pop balloons.
I am planning to document this project and make a nice video. At the end, I hope to publish the complete specs and code so the project can be replicated quickly. I have also been talking with a few profs and researchers in the field to see if there is interest in buying a kit that includes everything you need to fly a quadcopter with motion capture. It would reduce the time to start research on quadcopter from weeks or months to half a day, while staying affordable for many researchers. The platform could even be used in undergrad courses. Please contact me if you are interested.
Right now, I am testing mounting the balloons and conducting test with the laser to determine the time required to pop a balloon. I will do my best to update frequently. Please don't hesitate to comment and share your suggestions!
I did some tests. The lasers I ordered are for cutting or engraving. This means that their focal point is relativety narrow. I have to focus them, so set of points that will pop the balloon is a line about 15 cm long in front of the quadcopter. That should be good enough for now. I also have two relays that will be in series with the laser: one will be controlled from a manual wireless transmitter, and one will be controlled from the quad. This is for safety.
Just have to say... this is awesome... and scary at the same time... the technology that we have at our disposal is so powerful and cheap that it's scary to think what will happen with these devices in the next few years.
Stay safe, and keep the lasers pointed away from you!
Thanks! We are still working on this, but we were a bit delayed by other projects. Also, I switched to a PX4 integration with ROS. It makes a lot more sense in term of implementation. The PX4 is a lot more optimized for real time flight, and it has a good ROS integration.