Design of autonomous sustainable unmanned aerial vehicle - A novel approach to its dynamic wireless power transfer

Abstract

Electric UAVs are presently being used widely in civilian duties such as security, surveillance, and disaster relief. The use of Unmanned Aerial Vehicle (UAV) has increased dramatically over the past years in different areas/fields such as marines, mountains, wild environments. Nowadays, there are many electric UAVs development with fast computational speed and autonomous flying has been a reality by fusing many sensors such as camera tracking sensor, obstacle avoiding sensor, radar sensor, etc. But there is one main problem still not able to overcome which is power requirement for continuous autonomous operation. When the operation needs more power, but batteries can only give for 20 to 30 mins of flight time. These types of system are not reliable for long term civilian operation because we need to recharge or replace batteries by landing the craft every time when we want to continue the operation. The large batteries also take more loads on the UAV which is also not a reliable system. To eliminate these obstacles, there should a recharging wireless power station in ground which can transmit power to these small UAVs wirelessly for long term operation. There will be camera attached in the drone to detect and hover above the Wireless Power Transfer device which got receiving and transmitting station can be use with deep learning and sensor fusion techniques for more reliable flight operations. This thesis explores the use of dynamic wireless power to transfer energy using novel rotating WPT charging technique to the UAV with improved range, endurance, and average speed by giving extra hours in the air. The hypothesis that was created has a broad application beyond UAVs. The drone autonomous charging was mostly done by detecting a rotating WPT receiver connected to main power outlet that served as a recharging platform using deep neural vision capabilities. It was the purpose of the thesis to provide an alternative to traditional self-charging systems that relies purely on static WPT method and requires little distance between the vehicle and receiver. When the UAV camera detect the WPT receiving station, it will try to align and hover using onboard sensors for best power transfer efficiency. Since this strategy relied on traditional automatic drone landing technique, but the target is rotating all the time which needs smart approaches like deep learning and sensor fusion. The simulation environment was created and tested using robot operating system on a Linux operating system using a model of the custom-made drone. Experiments on the charging of the drone confirmed that the intelligent dynamic wireless power transfer (DWPT) method worked successfully while flying on air.