bjkomer / masters-thesis Goto Github PK

Hi Brent,

I am a Masters Electrical & Mechanical Engineering student at the University of Strathclyde, and I am trying to implement a neural adaptive flight controller in a physical drone. I have found your thesis very helpful and I want to see if I can call your different adaptive controllers in a similar way but instead of using V-rep I will be interfacing them in a drone I have built. The input drone state will come from IMU and GPS data and I will change the controller output from rotor speeds to PWM.

I have managed to get your simple_quadcopter model that uses the nengo GUI working with V-rep EDU 3.2.1 but when I have tried to use your run_model script and call any of your adaptive controllers from your masters-thesis I haven't been able to get them working. It seems that they only start to work when I slow the simulation down to 1ms in V-rep but they are still fairly unstable, I am wondering could this be due to the controllers not being up to date with the changes in nengo?

Many thanks,
Adam

Hello,I am now working at a quadcopter with a hand. In simple demo controller code given by vrep,it uses 9 as gravity compensation. But after I modified gravity_compensation in crontoller.py, it still can't run stably. Also, None of the adaptive ways worked.
What's needed to be modified apply this controller in my quadcopter?
Thanks.