| With the development of unmanned aerial vehicle(Unmanned Aerial Vehicle,UAV)technology,the use of UAVs for transporting has become an area of concern.There are mainly three ways to carry loads using quadcopter UAVs: carrying loads with rigid connections,suspending loads with ropes,and carrying loads with a mechanical arm.Among them,the use of suspended ropes has good adaptability and expansibility,but the swinging of the suspended load during transportation increases the instability of the entire transportation system and affects the flight safety of the UAV.To address this issue,this paper proposes a UAV-suspended load transport minimum swing trajectory generation system based on reinforcement learning.The main research contents are as follows:Firstly,the system model of a quadcopter UAV was established in this paper,and then the kinematic and dynamic models of the quadcopter UAV-suspended payload system were built based on this model.The model was partially simplified,such as assuming that the suspension rope is always taut and there is no wind indoors.The motion of the suspended payload was treated as a conical pendulum motion,and the kinematic equation of the system was established based on this.The dynamic equation of the system was established using the Lagrangian equation.Finally,the model can predict the next state of the quadcopter UAV-suspended payload system based on its current state and the control input of the UAV.Secondly,a region approximation exploration Deep Q-Networks(Deep Q-Networks,DQN)algorithm based on reinforcement learning is proposed for the quadcopter UAV-suspended load transport system.The algorithm takes the current state and action of the system as input to the neural network and selects the action with the maximum Q value as the next action,ultimately generating a trajectory with minimal swinging of the suspended payload.A simulation space compliant with the gym inputoutput standard was established based on the model of the quadcopter UAV-suspended payload system,and a suitable reward function was designed for the task the system needs to complete.Finally,the effectiveness of the designed algorithm and reward function was demonstrated through simulation.Finally,to further verify the effectiveness of the algorithm and the trajectory generated,a quadcopter UAV-suspended load system experimental platform is assembled.The UAV platform adopts a Z450 frame,equipped with a PixaHwk2.4.8flight control and a JetsonXavierNX onboard computer.The onboard computer runs Ubuntu 20.04 and ROS.The Opti Track motion capture system is used as the positioning source,and communication with the UAV is achieved via WiFi.The quadcopter UAV was flown with a single load weight and double load weight suspended,and flight experiments were conducted with the system generating the trajectory with minimal swinging and only given the target point.The results showed that the quadcopter UAV-suspended payload minimum swing trajectory generation system designed in this paper can effectively reduce the swinging of the suspended payload during flight. |
