Research On Key Technologies Of Combat Mission Planning For Manned/unmanned Units

With the development of unmanned combat equipment,manned/unmanned combat as a new form of combat has received extensive attention from many research institutions.In recent years,the proportion of unmanned systems in cooperative operations has increased,coupled with the improvement of intelligence levels has begun to change the style of cooperative operations,making the current manned-based operational command technology difficult to adapt to the characteristics of the cooperative operations,especially in the mission planning and decision-making.As an important component of command and control in combat,mission planning and decision-making will greatly affect the effectiveness of combat.This paper takes manned/unmanned units in small-scale battles as the characteristics of collaborative operations and conducts in-depth research around the task allocation at the battle level,and the guidance path generation in general marching process for manned/unmanned units.The main work accomplished in this paper is as follows:(1)Aiming at the task allocation and guidance path generation in small-scale battles,in this paper,the technical requirements of manned/unmanned units in terms of task allocation and route planning in conjunction are analysed,and suitable basic algorithms for task allocation technology and route planning technology are selected respectively,and then the corresponding specific research contents are determined.Further,a system architecture integrating mission allocation and route planning is proposed.(2)To realize task assignment decisions for the manned/unmanned units in small-scale battles,a multi-level task order generation network model based on an encoding-decoding framework is proposed,due to the presence of the fog of war and the large set of task assignment orders for manned/unmanned units.Aims to address the lack of explicit adversary strategies in network training and the need of diverse strategies for manned/unmanned units in small-scale battles,a dual coalition pool multi-intelligence training method based on the asynchronous advantage actor-critic(A3C)algorithm is proposed.Further,based on the proposed method,an implementation model with reinforcement learning combined with imitation learning is designed,resulting in a set of intelligent decision-making techniques that support the real-time generation of centralized task allocation decisions.AI development experiments on the wargame platform validate that the technique can effectively make centralized task allocation decisions in real-time for manned/unmanned units in a small-scale battle.(3)A study of single-machine static route planning is developed,using the artificial potential field(APF)method as the base algorithm.The refined construction method of planning space based on simple circular modeling is proposed.To address the problem of unnecessary collision avoidance at APF iterative path points,a method is proposed to dynamically adjust the effective influence distance of obstacles based on a collision hazard assessment.A trap-free route planning method based on the virtual obstacle method is proposed for the problem of local minimal value trap existing in the virtual potential field of APF.Further,an implementation model for single-engine static route planning based on APF is first proposed,and then an implementation model for single-engine static route planning capable of coping with the effects of the details from complex form barriers is formed based on this implementation model for a finely modeled planning space.Simulation experiments verify that the proposed method and model can effectively cope with the above problems.(4)Based on the static route planning technology of a single machine,the dynamic route planning technology is investigated.A new temporal collision risk assessment model is proposed for the dynamic adjustment of goals non reachable with obstacles nearby(GNRON).A virtual obstacle method is proposed for the trap problem in dynamic planning space with local minima.A cluster dynamic route planning method based on a dual potential energy field is proposed for a cluster composed of some units.To enable the planned path to guide cluster units,an orderly travel guidance method is proposed.Further,an APF-based single-unit dynamic route planning algorithm implementation model is first proposed for the general travel process.Based on this,a cluster dynamic route planning algorithm with configuration constraints is proposed for unmanned clusters.Simulation experiments verify that the proposed method and model can effectively solve the above problems.(5)Example demonstration and analysis.A small-scale offensive and defensive battle between red and blue ground combat units is the scenario with the help of the "Temple Calculator" military chess rehearsal platform.It is analyzed that the guidance path generation technology proposed in this article for the general marching process in manned/unmanned unit operations is feasible at the application level.AI developed by using the intelligent decision technique proposed in this paper for centralized task allocation at the battle level can learn human tactics to a certain extent,and apply these strategies to the war game relatively well.