A high-fidelity simulation test-bed for fault-tolerant octo-rotor control using reinforcement learning

A framework for fault-tolerant control of an octo-rotor unmanned aerial vehicle (UAV) is presented. UAVs have seen a sharp rise in use in recreational and industrial communities, and governmental organizations. Octo-rotors provide additional redundancy with motors and have the capacity to lift larger loads lending themselves to many performance-critical applications. Recent work on autonomous control of UAVs has primarily focused on quad-rotors. This work contributes to the body of research by presenting a simulator to test adaptive control under faults for an octo-rotor test-bed using reinforcement learning approaches.