With the rapid development of unmanned aerial vehicle(UAV)technology in China

UAVs are widely applied to low-altitude mapping

environmental monitoring

and urban planning.However

autonomous obstacle avoidance of UAVs in complex environments still faces challenges

as traditional path planning methods hardly meet realtime and intelligent requirements.An improved algorithm

proximal policy optimizationgoal object(PPO-GO)

based on the proximal policy optimization(PPO)

is proposed to address the issues of low training efficiency and insufficient sample utilization commonly observed in traditional reinforcement learning methods.The proposed algorithm introduces a clipping function during each policy update to prevent the model from falling into local optima.Meanwhile

simulation experiments are completed on the AirSim platform via the hierarchical training method.Experimental results demonstrate that the proposed method achieves faster convergence and a higher success rate in dynamic and complex obstacle scenarios.Compared with traditional methods

it enhances autonomous decision-making capabilities in complex environments

reduces the computational overhead of path planning

and provides practical and feasible solutions for low-altitude surveying

urban modeling

and emergency rescue applications.