The Design of Vehicle Longitudinal Collision Early Warning System Based on DSP and ARM9

Using image sensors to perceive the road traffic environment and the location of obstacles ahead, to achieve safe distance measurement, and to alert vehicles in danger of collision in time is beneficial to reduce traffic accidents and improve road traffic safety. Since the theoretically calculated safe distance must first be based on ensuring safety, it often differs from the safe distance recognized by the driver during driving, which leads to the driver’s distrust of the early warning system, which is not conducive to the promotion of the system use. At the same time, as the processing platform of the safety assisted driving system, the volume, cost, and functional redundancy of the PC are bottlenecks that are difficult to overcome in the application of on-board systems.

introduction

Using image sensors to perceive the road traffic environment and the location of obstacles ahead, to achieve safe distance measurement, and to alert vehicles in danger of collision in time is beneficial to reduce traffic accidents and improve road traffic safety. Since the theoretically calculated safe distance must first be based on ensuring safety, it often differs from the safe distance recognized by the driver during driving, which leads to the driver’s distrust of the early warning system, which is not conducive to the promotion of the system use. At the same time, as the processing platform of the safety assisted driving system, the volume, cost, and functional redundancy of the PC are bottlenecks that are difficult to overcome in the application of on-board systems.

Based on the image method of measuring the distance between the vehicle and the preceding vehicle, this paper establishes a vehicle longitudinal collision warning model to solve the contradiction between the theoretically calculated safety distance and the habitual distance recognized by the driver; consider the real-time processing of the embedded system With the characteristics of small size and so on, the embedded method is used to complete the design of the vehicle longitudinal collision warning system.

1 Measurement of the distance between vehicles ahead

The image sensor is fixed on the top of the front of the vehicle, and the height of the camera from the ground is h. The establishment of the spatial coordinate system is as follows:

The Design of Vehicle Longitudinal Collision Early Warning System Based on DSP and ARM9

The Design of Vehicle Longitudinal Collision Early Warning System Based on DSP and ARM9

The world coordinate system XYZ and the camera coordinate system xyz. The dynamic world coordinate system moves with the vehicle. The vertical projection point of the center of the camera lens on the ground is the origin. The vertical line of the ground upward is the positive direction of the Z axis, the direction of the longitudinal axis of the vehicle body is the X axis, and the positive direction is the opposite direction of the car. ; The camera coordinate system takes the optical axis as the z axis, the lens center as the coordinate origin, and the camera coordinate xy plane is parallel to the image plane; the u axis and v axis of the image plane coordinate system representing the position of each point in the image are parallel to the x axis and y axis , The origin is at the center of the image, which is the intersection of the camera’s optical axis and the image. All coordinate systems satisfy the right-hand rule.

The measurement of the vehicle or obstacle in front by the image sensor includes distance measurement using a single frame image and distance measurement using multiple frames of images. The installation position parameters of the camera are considered in the research. The angle between the z coordinate axis of the camera and the X axis is called the sweep angle β, and the angle between the optical axis of the camera (z axis) and the vertical direction (Z axis) is called the tilt angle α. As shown in Figure 1. In the figure, the camera’s external parameters α, β, h and internal parameter camera focal length f are obtained by strict camera calibration. The vehicle’s image characteristics and Kalmam filter principle are used to realize vehicle identification, and the computer image coordinates of point P at the bottom of the vehicle can be known.The point (u0, v0) is the origin of the image plane coordinates (O, 0); the point (u, v) can be calculated by the P point computer image coordinates (m, n) according to the internal parameter model formula

The Design of Vehicle Longitudinal Collision Early Warning System Based on DSP and ARM9

get. Among them, kx, ky, are the magnification factors of the digital image in the x-axis and y-axis directions; Om, On are the computer image coordinates of the origin of the image plane. According to the perspective projection and the triangle geometric relationship, the distance between the two cars is calculated by formula (1):

The Design of Vehicle Longitudinal Collision Early Warning System Based on DSP and ARM9

In order to verify the calculation accuracy of formula (1), after the camera calibration is completed, the crosswalk of the road is photographed, the horizontal distance between each crosswalk and the center of the camera lens is measured, and the distance is compared with the distance calculated according to the distance measurement model. The results are shown in Table 1. Shown. In the table, the error between the measured distance and the calculated distance includes measurement error and model error. It can be seen from the table that the error is within the allowable range and can meet the processing requirements for the next step.

2 Determination of the driving state of the preceding vehicle

The driving state of the preceding vehicle affects the pre-warning time of the vehicle longitudinal collision warning model. The judgment of the preceding vehicle state is based on the relative distance and relative speed of the vehicle ahead. According to the image sequence frame, the distance between the current time and the next time between the vehicle and the vehicle in front is measured, and the instantaneous speed of the current time and the next time is obtained through the speed sensor of the vehicle, then:

The Design of Vehicle Longitudinal Collision Early Warning System Based on DSP and ARM9

Where: L2, L1, L0 are the distances between the vehicle and the preceding vehicle measured at different times (unit: m); vb1, vb2, vq1, vq2 are the speeds of the vehicle and the preceding vehicle at different times (unit: m/ s); vrel1, vrel2 are the relative speeds of the vehicle and the preceding vehicle at different times; ab and aq are the decelerations of the vehicle and the preceding vehicle at that moment (unit: m/s2); △t is the interval time (unit : S).

(1)|vq1, vq2|
(2)|vq1, vq2|>εv, and |vq1-vq2|
(3)|vq1, vq2|>εv, and |vq2-vq1|>εv, and vq2The Design of Vehicle Longitudinal Collision Early Warning System Based on DSP and ARM9

3 Longitudinal collision warning algorithm based on the state of the vehicle ahead and the safety factor

3.1 Establishment of a longitudinal collision warning model

While ensuring driving safety and highway traffic capacity, the vehicle longitudinal collision warning system must also maintain the driver’s trust in the system. If the warning safety distance of the early warning system is often greater than the driver’s own judgment of the safety distance, The frequent alarms of the system may cause the driver to ignore the alarm signal of the system or give up the use of the system. The early warning algorithm designed in this paper uses the state of the preceding vehicle to determine the safety factor, improves the system’s early warning timing control, improves the safety performance of the system’s early warning, and increases the trustworthiness of the system.

The establishment of the minimum safe distance warning algorithm is as follows:

The Design of Vehicle Longitudinal Collision Early Warning System Based on DSP and ARM9

In the formula: Ld is the safe distance when the early warning system starts to alarm; Ls is the minimum safe distance calculated according to the vehicle braking theory and when the vehicle is in a different state from the preceding vehicle; γ is the weight of the safety factor based on the state of the preceding vehicle. The formula for calculating the minimum safe distance Ls is as follows:

The Design of Vehicle Longitudinal Collision Early Warning System Based on DSP and ARM9

In the formula: t is the reaction time of braking operation (unit: s); vs, vq are the initial speeds of the car and the preceding vehicle before braking (unit: m/s); vrel is the relative initial speed of the two cars (unit: m/s-); D0 is the safe distance when the two cars stop or when the speeds of the two cars are equal, generally 2~5m.

The value rules of the safety factor γ are as follows:

The Design of Vehicle Longitudinal Collision Early Warning System Based on DSP and ARM9

The vehicle in front is at a standstill, or the vehicle in front is driving at a constant speed and the speed of the vehicle in front is faster than that of the vehicle in front, that is, aq=0, the safe distance calculated based on the maximum deceleration of the vehicle is more reasonable. In actual traffic, the driver considers the comfort of the car and seldom uses the maximum deceleration operation, and has taken measures before reaching the minimum safe distance. The main function of the early warning model is to remind the driver of the operating state when the driver is negligent or distracted , So the safety factor weight γ=1.

When the vehicle in front suddenly decelerates, there are three situations:

(1) The decelerations of the two vehicles are equal, and the general conditions for highway driving are met. Based on the safety distance calculated in this case, the safety factor weight γ is determined by measuring the deceleration of the preceding vehicle.

(2) The deceleration of the vehicle is less than the deceleration of the vehicle in front, and the braking efficiency of the vehicle is inferior to that of the vehicle in front, which is the most dangerous situation, but the probability of occurrence is small.

(3) The deceleration of the vehicle is greater than the deceleration of the vehicle in front, the braking intensity of the vehicle is higher than that of the vehicle in front, or the braking intensity of the vehicle changes with the change of the vehicle in front, and is always higher than that of the vehicle in front. Common situation. At this time, the calculated safety distance is too small to be used as an early warning basis.

In summary, when aq=0, γ=1 as described above; when aq3.2 is verified by the test

The Wuling Zhiguang 6400C3 extended version mini-car was selected on the structured road as the experimental vehicle, and the above-mentioned safe distance warning algorithm based on the state of the preceding vehicle and the safety factor was used to verify the warning time of the vehicle following distance. The results show that the system can reliably give early warning, and the distance between the workshops at the time of warning is acceptable to the driver.

Figure 2 is a screenshot of the alarm moment when the vehicle in front is driving at a constant speed and the vehicle is accelerating. At this time, the distance between the workshops is 24.895 m, and the relative speed of the two vehicles is 5.513m/S;

The Design of Vehicle Longitudinal Collision Early Warning System Based on DSP and ARM9

Figure 3 is a screenshot of the alarm moment when the vehicle in front is driving at a reduced speed and the vehicle is driving at a constant speed of 90 km/h. At this time, the distance between the workshops is 45.847m, and the relative speed of the two vehicles is 8.571m/s.

The Design of Vehicle Longitudinal Collision Early Warning System Based on DSP and ARM9

4 System design

The real-time performance of the automobile longitudinal collision system requires the processor to have a faster running speed and a strong real-time scheduling capability. In the study, the Davinci technology (Davinci) designed and produced by Texas Instruments (TI), based on DSP and The evaluation board of the ARM9 dual-core TMS320DM6446ZWT system-on-chip (SoC) is used as the hardware platform of the system, and the Linux system is selected as the embedded operating system.

The Design of Vehicle Longitudinal Collision Early Warning System Based on DSP and ARM9

The system first receives the image transmitted by the image sensor, converts the color image into a gray image to be processed, and uses image processing methods such as median filtering, Sobel operator edge detection, and adaptive threshold segmentation to eliminate noise and smooth the image. Edge detection and image segmentation are used to obtain binarized images.

Recognize the outline of the vehicle in front on the basis of the binary image and gray image, determine the outline size of the vehicle in front and the position of its bottom edge in the plane image, and realize the calculation of the following distance and the safe distance according to the longitudinal collision warning model of the vehicle Early warning. The software flow chart is shown as in Fig. 4.

5 Conclusion

A vehicle longitudinal collision early warning algorithm based on the motion state of the preceding vehicle and the weight of the safety factor is proposed to ensure driving safety while ensuring highway traffic capacity. The calculated alarm distance is consistent with the car-following safety distance recognized by the driver, which improves the performance of the system. Trust degree; the application of embedded system effectively reduces the volume of the system, which is conducive to the application and promotion of the system.

The Links:   6DI20C-050 G154I1-LE1

Related Posts

Leave a Reply

Your email address will not be published. Required fields are marked *