Method for path planning of equipped vehicle based on transfer learning

Abstract

The invention belongs to the technical field of vehicle path planning, and in particular relates to a method for path planning of an equipped vehicle based on transfer learning. The method comprises the following steps of S1, obtaining basic data of the equipped vehicle; S2, obtaining planning time and a planning goal; S3, obtaining static planning environment data; S4, obtaining equipped vehicledriving data; S5, constructing a path planning model by using a DDPG (Deep Deterministic Policy Gradient) algorithm; S7, obtaining parameter variation data of a dynamic planning environment; S8, constructing a dynamic planning environment domain; S9, slightly adjusting parameters of a deep neural network; S10, with the trained network parameters as input of a path planning algorithm, obtaining theterrain and intelligence data in a battle in real time, and constantly adjusting planning strategies to generate a route planning result for the equipped vehicle; and S11, obtaining the path planningresults of different battlefield environments through the dynamic planning environments in different battles, and taking the path planning results and the corresponding network training parameters thereof as historical samples.

Description

technical field [0001] The invention belongs to the technical field of vehicle path planning, in particular to a method for path planning of equipped vehicles based on transfer learning. Background technique [0002] The path planning problem of equipped vehicles has been studied by scholars. From simple static road network planning to dynamic planning in complex scenarios, the model algorithm of path planning has been continuously improved and upgraded. The path planning of the equipped vehicle not only considers the inherent constraints of the equipped vehicle, including the length, width, load-bearing, turning radius, wading ability, obstacle-surpassing ability, and climbing ability of the equipped vehicle, but also considers complex scenarios, especially in wartime. The constraints of terrain and meteorological environment on the path planning problem require a variety of factors and matching parameters to control the output of a reasonable path. The DDPG algorithm base...

AI Technical Summary

Problems solved by technology

[0004] In the actual combat maneuver planning system, the battlefield environment faced is mostly a dynamically changing environment. Not only the terrain in the war has a great impact on the driving of equipped vehicles, but also the establishment and destruction of enemy fire blockade areas and nuclear, biological and chemical contaminated areas will cause equipment vehicles to Possibility of diversion

At this time, the static DDPG algorithm takes a lot of time due to the pre-training, and it is difficult to meet the real-time requirements of sudden changes in the situation and changes in the plan.

Images