Using Hector Quadrotor with gazebo and a neural network

asked 2016-12-19 17:51:34 -0500

graffy gravatar image


I'm faced with trying to make a UAV fly in gazebo using machine learning. The project is a bit ambiguous, and I'm pretty unfamiliar with Gazebo / robotics / UAVs apart from a passing general knowledge.

That said, I've spent some time working with Gazebo and the tutorials. Writing plugins to do the job doesn't seem all that difficult, and I took a look at the controller code for the hector quadrotor for a clue on how to build that part.

That said, I still really have no idea how to get Hector working in gazebo using plugins (no ROS). I see that the source was built with ROS in mind, and I also see that the model/sensor files are of the type urdf.xarco, which is not the same as an sdf.

Maybe I don't really need Hector to do the job - with sophisticated enough controller code, I'm sure the simple quadrotor model in gazebo would be just fine. Still, it would be nice if there was a more direct example of running simulations in gazebo without ROS... Even better if it demonstrated how to drive a gazebo sim with an AI...

Anyway, grateful for any tips or suggestions...


Anyway, I'm continuing to dig and follow leads, but I was hoping maybe someone could point me to something helpful...

edit retag flag offensive close merge delete