阅读说明：本技术 机动车保护行人的方法 (Method for protecting pedestrians for motor vehicle ) 是由 齐杰 马科 于 2021-07-08 设计创作，主要内容包括：本发明公开了一种机动车保护行人的方法,利用双目摄像头采集行人的世界坐标,输入到生成性对抗网络的社会可接受轨迹模型转换为物理坐标,并预测连续的行人物理坐标,通过相邻坐标的比较计算行人运动方向,用双目摄像头测量预测的行人物理坐标中行人与车辆的距离,当预测最近距离小于阈值时,弹出发动机舱盖下和前保险杠处的行人安全气囊；本发明能够快速准确预测行人的运动轨迹,当预测到行人与车辆即将碰撞时弹出安全气囊,能够较好地保护行人安全,克服了雷达测距容易误触发的缺陷。(The invention discloses a method for protecting pedestrians by a motor vehicle, which comprises the steps of collecting world coordinates of pedestrians by using a binocular camera, inputting the world coordinates into a socially acceptable track model of a generative countermeasure network, converting the world coordinates into physical coordinates, predicting continuous physical coordinates of the pedestrians, calculating the moving direction of the pedestrians by comparing adjacent coordinates, measuring the distance between the pedestrians and vehicles in the predicted physical coordinates of the pedestrians by using the binocular camera, and popping out pedestrian airbags under an engine compartment cover and at a front bumper when the predicted closest distance is smaller than a threshold value; the invention can quickly and accurately predict the motion trail of the pedestrian, and the safety airbag is popped up when the imminent collision between the pedestrian and the vehicle is predicted, so that the safety of the pedestrian can be better protected, and the defect that radar ranging is easy to trigger by mistake is overcome.)

1. A method of protecting a pedestrian in a motor vehicle, comprising the steps of: (1) a pedestrian safety airbag is arranged under an engine compartment cover and at a front bumper of the automobile, and a binocular camera is arranged at the upper part of a front window of the automobile; (2) the binocular camera detects the distance between a pedestrian and a vehicle in real time and predicts the pedestrian track; (3) and (3) if the predicted result in the step (2) is that collision is possible, bouncing off the pedestrian airbag.

2. The pedestrian protection method for motor vehicles according to claim 1, wherein in the step (2), the method for predicting the pedestrian trajectory is: a world coordinate system is established by taking a binocular camera as an origin, coordinate points of pedestrians are continuously collected, physical coordinates of the pedestrians are predicted by utilizing a network-generated complex real acceptable track model, and the moving direction and the predicted closest distance of the pedestrians are calculated.

3. The pedestrian protection method of claim 2, wherein the binocular camera detects the world coordinates of the pedestrian once every 400ms, continuously detects 8 coordinate points, generates a network complex reality acceptable trajectory model, predicts using the 8 coordinate points, and outputs the physical coordinates of the 8 pedestrians at time intervals.

4. The method of protecting pedestrians for motor vehicles according to claim 2, wherein the generating network complex reality acceptable trajectory model subtracts the vehicle advance distance from the predicted pedestrian distance to obtain predicted physical coordinates of the pedestrian.

5. A method for protecting a pedestrian according to claim 2, wherein the direction of movement of the pedestrian is obtained by comparing two adjacent coordinate points in the physical coordinates.

6. The method for protecting pedestrians by motor vehicles according to claim 2, characterized in that the physical coordinates are measured by a binocular camera, and the predicted closest distance to the pedestrian is screened out.

Technical Field

The invention relates to a method for protecting pedestrians for a motor vehicle.

Background

The pedestrian protection safety airbag can effectively prevent a human body from impacting the front windshield of the automobile so as to prevent pedestrians and passengers in the automobile from being more injured under violent collision, and particularly, the most common pedestrian casualty accidents can be reduced by using the engine hood airbag and the front wall safety airbag in a matched manner; the existing airbag triggering mechanism is mainly completed based on a radar sensor, the radar sensor has the defect that pedestrians and obstacles cannot be effectively distinguished, the triggering condition is generally set to be a distance, and when the obstacles approach or a vehicle owner is close to the obstacles in distance during parking, false triggering is easily caused.

Disclosure of Invention

The purpose of the invention is as follows: the invention aims to provide a pedestrian protection method for triggering an air bag according to a prediction result of a pedestrian track and a distance between the pedestrian track and a vehicle, so that the safety protection effect is improved, and the false triggering probability is reduced.

The technical scheme is as follows: the method for protecting pedestrians by the motor vehicle comprises the following steps: (1) a pedestrian safety airbag is arranged under an engine compartment cover and at a front bumper of the automobile, and a binocular camera is arranged at the upper part of a front window of the automobile; (2) the binocular camera detects the distance between a pedestrian and a vehicle in real time and predicts the pedestrian track; (3) and (3) if the predicted result in the step (2) is that collision is possible, bouncing off the pedestrian airbag.

A world coordinate system is established by taking a binocular camera as an origin, the world coordinates of pedestrians are continuously collected and input into a Social GAN (Social Acceptable resources with a generated comprehensive Acceptable track) model, the physical coordinates of the pedestrians are conveniently and quickly generated, the pedestrian track is accurately predicted, the moving direction and the predicted closest distance of the pedestrians are calculated, when the predicted closest distance of the pedestrians is smaller than a threshold value, an engine compartment cover automatically pops open, and pedestrian airbags under the engine compartment cover and at a front bumper are arranged.

Has the advantages that: compared with the prior art, the invention has the following advantages: the pedestrian distance can be accurately measured, the world coordinate system is quickly converted into the physical coordinate system, the pedestrian track is predicted, and the safety airbag is ejected to protect the safety of the pedestrian when the pedestrian is about to approach according to the prediction result.

Drawings

FIG. 1 is a flow chart of a pedestrian protection method of the present invention;

FIG. 2 is a physical coordinate system of a pedestrian of the present invention;

fig. 3 is a distance between a pedestrian and a vehicle measured using a ruler and the method of the present invention.

Detailed Description

The technical scheme of the invention is further explained by combining the attached drawings.

As shown in fig. 1, the method for protecting pedestrians of a motor vehicle of the present invention includes the steps of:

(1) pedestrian safety airbags are arranged under an engine compartment cover and at a front bumper, the safety bags under the engine compartment cover are used for protecting the head of a collided pedestrian, and the airbags at the front bumper are used for protecting the knees of the pedestrian.

(2) The method comprises the following steps of installing a binocular camera on the upper portion of a front window of an automobile, shooting left and right images of pedestrians in real time by the binocular camera and predicting tracks of the pedestrians.

(2.1) As shown in FIG. 2, a world coordinate system is established with a binocular camera as an origin, and a world coordinate P (X) of a pedestrian is detected every 400ms_w,Y_w,Z_w) Continuously detecting 8 points, predicting the pedestrian motion track of the 8 points by using a Social GAN model, and generating the actual physical coordinate P '(X') of the pedestrian_c,Y_c,Z_c) And outputs the predicted physical coordinates (X) of 8 points at intervals of time_c1,Y_c1,Z_c1)，……(X_c8,Y_c8,Z_c8). Because the coordinate system is changed in the advancing process of the vehicle, even if the coordinate system is changed without moving a person, the person is mistakenly considered to advance, so that wrong prediction is made, and in order to solve the problem, the output Y_cThe value is subtracted by the vehicle advancing distance to eliminate errors, the distance of the pedestrian is guaranteed to be more accurate, the vehicle advancing distance is provided by the vehicle control module, and the calculation formula is as follows:

S₁＝S₂+0.01*0.073*π*v

where v is the wheel speed S₁Is the actual distance, S₂For the test distance, the engine speed ratio was 9.7, the damping number was 60, the air-fuel ratio was 0.703, and the error coefficient was 0.01.

And (2.2) comparing the coordinates of two adjacent points in the 8 points which output the prediction, and if the coordinate change directions are consistent as the result of at least four times of comparison, judging that the moving direction of the pedestrian is the same direction.

(2.3) ranging the first 5 points of the 8 points which output the prediction, determining the distance between the pedestrian and the binocular camera through the parallax of the left image and the right image of the pedestrian, and screening out the predicted closest distance of the pedestrian.

(3) Setting the distance between the vehicle head and the binocular camera as X, and if the predicted pedestrian closest distance is more than X, not popping up a pedestrian safety airbag by the vehicle; and predicting that the shortest distance of the pedestrian is less than or equal to X, automatically ejecting the engine compartment cover, and ejecting pedestrian safety airbags below the engine compartment cover and at the front bumper.

The distance between the pedestrian and the vehicle is measured by using a ruler and the method respectively, the following data are obtained, and as shown in figure 3, the distance measurement precision of the binocular camera within 10m is very high and reaches the cm level.