Unmanned aerial vehicle obstacle avoidance method based on deep reinforcement learning

Abstract

The invention discloses an unmanned aerial vehicle obstacle avoidance method based on deep reinforcement learning, and the method comprises the following steps: 1), building an unmanned aerial vehicle obstacle avoidance flight model in a three-dimensional space, and randomly generating the number and position of obstacles, and the starting point of an unmanned aerial vehicle; (2) establishing an environment model based on a Markov process framework, (3) selecting actions based on states and strategies, enabling the unmanned aerial vehicle to interact with the environment to generate a new state after taking the actions, calculating an obtained reward, forming quaternions by the states, the actions, the reward and the actions at the next moment, and storing the quaternions in a sample space through an improved method for sample sampling training; 4) performing network updating on a sample obtained by sampling the environment model by adopting an improved DDQN algorithm, and assigning a state-action pair of the sample; and 5) selecting an optimal action according to the assignment of each action in the state in the sample, and further obtaining an optimal strategy. The invention provides the reinforcement learning obstacle avoidance method adopting the segmentation sampling pool, and the training efficiency of the generation strategy is improved.

Description

technical field [0001] The invention belongs to the technical field of intelligent decision-making, and in particular relates to an obstacle avoidance method for unmanned aerial vehicles based on deep reinforcement learning. Background technique [0002] As UAVs play an increasingly important role in military warfare and civilian fields, UAVs are required to fly autonomously to complete tasks without human intervention in a variety of mission scenarios. Therefore, finding a suitable method to solve the problem of obstacle avoidance when UAVs fly autonomously can improve the success rate of UAV missions to a certain extent. Traditional obstacle avoidance methods, such as artificial potential field method, visual graph method, and particle swarm optimization algorithm, are very mature, but they need to establish different models according to different situations. However, in the actual flying environment of UAVs, UAVs are often required to detect and make real-time decisions ...

AI Technical Summary

Problems solved by technology

Reinforcement learning has been applied to the obstacle avoidance problem of drones. Since obstacle avoidance of drones is a problem in continuous space, it is necessary to combine neural networks to assign values ​​to each state-action pair, but learning combined with neural networks is prone to training problems. In case of unstable process

Images