Quadcopter Drone Train DDPG agent to follow Trajectory path

I would like to know :

>> Is there an Matlab Example showing How to train Quadcopter Drone with RL agent to follow Trajectory Path ?

>> If isn't, Is it possible to train Quadcopter Drone with DDPG agent ?

>> I have seen the example of Quadcopter Project on Matlab with PID tuning but not with RL agent like DDPG or others.

>> As I am engineering student and very new to Control Dynamic,

Can somebody please point me out that Quadcopter Drone to train RL agent to follow Trajectory path?

I would be really thankful and appreciated for any help advice and guidance.

[Bay Jay]

I hope this helps:

1. Get an RL example of DDPG used in MATLAB.

https://www.mathworks.com/help/reinforcement-learning/ug/quadruped-robot-locomotion-using-ddpg-gent.html

Run the training of that example and read through how a reward function is created. Reward function is done used the parameters of the vehicle/quadrotor example yaw, positions

2. Obtain an Quadotor module built in simulink. Or you can build the simulink version follow the mathematical equations you know. Verify that it behavior like the mathematical model when coding in MATLAB. (Give an input and run it as an ODE)

2. In your application, the trajectory becomes like a reference that changes in time. Use MATLAB signal editor to creat the trajectory. Lets say you have trajectory for x,y. Then you can create the x(t) for one signal editor and y(t) for another signal edtion. And the total time is 10 second to completely track the trajectory. As the RL agent learns the current x and current y are fed to the RL agent as states/observations.

3. The reward function whoud be like an error of the current x(t) and y(t) compared to the Quadrotor values x,y

Start simple and learn as you progress. All the best.