Missing target searching method based on reinforcement learning algorithm

Abstract

The invention provides a missing target searching method based on a reinforcement learning algorithm. The missing target searching method comprises the following steps: S1, data preprocessing including: discretizing time and space; discretizing a target moving track; standardizing the search difficulty in different time spaces; S2, constructing a reinforcement learning training environment: constructing the reinforcement learning training environment, wherein the training environment information comprises expected search costs of objects starting from different positions at different times atdifferent search moments and the probability of transferring to different positions at different search moments; S3, off-line training of a space-time search model: defining states and behaviors and performing adaptive optimization on the model; and S4, online space-time search decision making: iteratively determining a space-time search sequence by adopting a greedy strategy based on the space-time search model trained in the step S3, and executing space-time search. According to the method, the search cost for finding the position of the target at the target moment is effectively reduced, and the target search task under the search cost constraint is completed.

Description

Technical field [0001] The invention belongs to the field of missing target search under the constraint of group intelligence perception cost, and in particular relates to a missing target search method based on a reinforcement learning algorithm. Background technique [0002] Finding missing targets (for example, cars or people) in the city is essential for city safety management. For example, a suspicious car was identified at a certain point in the past. Instruction attached figure 1 Shows typical application scenarios. The serial number indicates the waypoint of the vehicle trajectory from 8:00 to 9:00 on a certain day, including the position of some key moments (light color). By determining the location of the suspicious vehicle at a specific moment (ie the target moment) and marking it on the map, and so on, until the approximate trajectory can be grasped. Soon after this, the police seized this clue and planned to monitor the suspicious vehicle's whereabouts to determin...

AI Technical Summary

Problems solved by technology

This presents a huge challenge for finding the target

Images