Humanoid intelligent controlled obstacle avoidance control method based on reinforcement learning

Abstract

The invention relates to a humanoid intelligent controlled obstacle avoidance control method based on reinforcement learning. The method comprises the following steps: S1, judging the distance from anobstacle to the vehicle, and determining the relation between the distance d and the safety distance S when the obstacle appears in the sensing area in front of the vehicle; if d < S, the vehicle cannot realize effective obstacle avoidance and would stop; if d >= S, and a relative speed exists between the vehicle and the obstacle, performing obstacle avoidance; if d >= S, and the relative speed is zero, the vehicle travels on the original route; S2, obstacle avoidance learning of the vehicle: calculating the turning radius and the accelerated speed of the vehicle when performing obstacle avoidance; S3, autonomous obstacle avoidance of the vehicle: after online learning of the vehicle, calling a driving experience to perform an autonomous obstacle avoidance by utilizing a memory function.For roads of various conditions, the obstacle avoidance control method provided by the invention has a relatively high obstacle avoidance rate, a good practicability and a good development potential.

Description

technical field [0001] The invention relates to the field of automatic driving, in particular to an obstacle avoidance control method based on human-like intelligent control of reinforcement learning. Background technique [0002] With the popularization of small smart cars, people have more choices in short-distance travel. At the same time, the application of smart cars in some special occasions has also played an increasingly important role, such as the execution of special tasks such as anti-terrorism and inspections. Therefore, research on assisted obstacle avoidance driving technology for narrow and complex roads is becoming more and more popular. Throughout the research on smart cars at home and abroad, most of them are biased toward unmanned vehicles. Unmanned vehicles have high costs based on multi-sensor technology and application limitations mainly on regular roads, so their core technologies are not fully applicable to small vehicles. smart car. [0003] In the...

AI Technical Summary

Problems solved by technology

Throughout the research on smart cars at home and abroad, most of them are biased toward unmanned vehicles. Unmanned vehicles have high costs based on multi-sensor technology and application limitations mainly on regular roads, so their core technologies are not fully applicable to small vehicles. smart car

[0003] In the past few decades, researchers have proposed many obstacle avoidance methods. For example, SS.GE and others have improved the gravitational and repulsive functions in the traditional artificial potential field method, and added the relative velocity parameters of obstacles and robots to the function. , to realize the effective obstacle avoidance of the mobile robot, but the controlled object is likely to fall into a local pole during the motion process and cannot reach the target point in the originally planned feasible path, and the computational complexity is very high, which makes it difficult to meet the real-time requirements; in addition, The reactive obstacle avoidance method has a good effect in a static environment, but it is not suitable for a dynamic environment

[0004] At present, the research on autonomous obstacle avoidance control of intelligent vehicles is mainly divided into two categories: traditional algorithms applied to static conditions and intelligent algorithms applied to dynamic scenes. Fully achieve real-time and fast autonomous obstacle avoidance effect

Images