Target Search in Unknown Environment Based on Temporal Differential Learning

This paper provides a path planning algorithm to solve the problem of target search in the unknown environment. By updating the value function after each action in turn, this paper overcomes the problem that traditional reinforcement learning algorithms require a large number of training processes. Besides, this paper further expands and optimizes the algorithm based on the hardware characteristics of UAV (Unmanned Aerial Vehicle). When the detection range of the sensor is different, the efficiency of the algorithm can be improved by taking it into consideration. To solve the problem of high steering time cost, it increase the number of possible non-existing paths based on the value function. The improvement and optimization for practical problems in this paper makes the algorithm can be applied to UAV better. Finally, the paper tests the algorithm in a simulation environment to ensure that the algorithm can effectively complete the path planning task of the search target.