Implementation of Quadrotor Model-based Learning in pytorch and VREP Simulator based of the following papers:

Low-Level control of a quadrotor with deep model-based reinforcement learning

Learning to adapt in dynamic, real-world environments through meta-reinforcement learning

We are testing separately Fault-Free Case, and Fault-Motor 1 case for same trajectory.

Here, we show trajectory followed by quadrotor in a Circle trajectory

Fault-Free Case, trajectory over time

Fault-M1 Case, trajectory over time

Same Comparison in 3D dim, Left Fault free, Right Fault Motor 1

Fault-Free Case, trajectory over time

Fault-M1 Case, trajectory over time

Same Comparison in 3D dim, Left Fault free, Right Fault Motor 1

In the following gif we show both in Point Trajectory, Left: Fault-Free Case, Right: Fault-M1 Case

Left Fault-free (Blue), Fault case (Yellow) in Helicoid & Vertical-Sin paths