| Definition | RL environments for controlling fixed-wing aircraft using the JSBSim flight dynamics model, integrated with the FlightGear flight simulator for realistic 3D visualization. |
| No of environments | 132 |
SBGym is a Python package designed to create and interact with reinforcement learning environments focused on the control of fixed-wing aircraft. Built upon the JSBSim flight dynamics model, SBGym provides a robust platform for researchers and developers to explore the complexities of aircraft control within a simulated environment.
Features of SBGym are supported by the FlightGear flight simulator, a widely recognized and open-source platform known for its realistic flight dynamics and extensive aircraft library. FlightGear enhances the SBGym experience by providing 3D visualization of flight scenarios, allowing users to observe the aircraft’s behavior in real time. With support for over 400 aircraft and a rich worldwide scenery database, FlightGear offers a highly configurable environment for simulating diverse flight conditions. This combination of SBGym and FlightGear enables an engaging platform for developing and testing reinforcement learning algorithms in the challenging domain of aviation.
| Example observations | Altitude, velocity etc. |
| Example actions | Actions are 3-tuples of floats in the range [-1,+1] describing commands to move the aircraft’s control surfaces (ailerons, elevator, rudder |
| Example reward | Fly in a straight line, maintain initial altitude and direction of travel. Additional reward for state transition. |