Adaptive round-up method for multi-robot hunters rounding up a single moving target

Abstract

The invention discloses an adaptive round-up device and method for multi-robot hunters to round up a single moving target, including the following methods: multi-stage round-up process: firstly, the round-up process is modeled in multiple stages, and the round-up task is divided into search and chase , encirclement, and capture four stages, and then adopt the strategies corresponding to the above four stages to control; the pursuit strategy based on the bionic neural network: the bionic neural network method adopted in the multi-robot system is migrated to the encirclement environment. Multi-robot hunters conduct hunting guidance; angle relationship encirclement method: use angle relationship to adjust the hunter's moving direction; multi-virtual potential energy point method: set multiple virtual potential energy points to form a capture formation to complete the final round-up task. The invention divides the round-up process into different stages, uses the bionic neural network method to solve the unknown, dynamic and real-time nature of the environment in the chasing stage, and completes the efficient obstacle avoidance chasing and round-up process.

Description

technical field [0001] The invention relates to the technical field related to robot chasing and rounding up, in particular to an adaptive rounding up device and method for multi-robot hunters rounding up a single moving target. Background technique [0002] Moving target siege means that the hunter surrounds the moving target, so that the moving target has no way to escape, and then takes the next step. This requires multi-hunters not to simply hunt independently, but to cooperate to complete the round-up task efficiently. Such collaboration among multiple pursuers is regarded as a robotic system in robotics, which becomes a challenging key problem and produces a lot of research results. The current research on the multi-robot round-up problem can be roughly divided into two models: the sensor-based model and the known localization model. In sensor-based models, a common control method is to capture in an unknown environment, guide and control by introducing sensor data. ...

AI Technical Summary

Problems solved by technology

However, there are still some limitations for these two models to be transferred to the multi-robot hunter problem: the sensor-based model only uses the data of the sensors carried, and does not use communication technology, making this method difficult, and the complexity of the algorithm varies with With the increase of the number of robots and evaders, it rises sharply; the known positioning model assumes that the position of evaders is known, which is very difficult in the real world, and then cannot satisfy the characteristics of simulating reality and being close to reality

There are also some methods that need to use training data sets for learning or data mining. When applied to specific multi-robot roundups, each change of environmental information such as maps requires different data sets, which is unrealistic and inflexible.

Images