A Learning-Capable Optimal Control Architecture for Energy Optimization in Hybrid Electric UAVs

The development of energy-efficient and low-cost propulsion systems for UAVs by implementing hybrid-electric concepts is critical for the US military. optoXense-GaTech team proposes to develop novel learning-capable optimal control methods to optimize the energy/power distribution for small UAVs with hybrid electric propulsion (HEP) systems for various flight scenarios. The method is based on robust optimal control and optimization approaches enabled with machine-learning tools. The control system will be developed to optimize the energy distribution and handle constraints, such as the case where the battery state of charge (SOC) reaches its boundaries. The benefits of such an energy optimization scheme, in addition to optimizing the energy/power/fuel consumption for different flight scenarios are: providing a decision-making tool for HEP configuration selection (i.e., series, parallel, etc.), and insight for integration of new high-energy devices onboard the UAVs.The initial configuration of the powertrain architecture will consist of a gearbox, electric motors, power electronics, batteries, a clutch and an internal combustion engine. In Phase-I, we will develop a control architecture to optimize energy/power distribution in small UAVs with HEP system, and conduct preliminary tests and verification of the algorithms at the software-level.In Phase-II, control hardware prototype for a small UAV will be demonstrated.