Research On The Departure Warning System Of Guardrail Cleaning Vehicle Based On Machine Vision

With the development of the city,the number of urban roads is increasing,and the urban road guardrails are also increasing.Road guardrails play a very i mportant role in improving the safety of road traffic and improving traffic order.In the work of the guardrail cleaning car,the driver needs to observe the distance between th e cleaning brush and the guardrail through the rear-view mirror while cleaning the car in the safe driving guardrail,and ensure the proper distance between the cleaning brush and the guardrail by controlling the vehicle.This way of working increases the driver's workload,making the driver easy to distract,driving,low cleaning ef ficiency,and potential safety hazards.This paper studies the detection and guardrail detection of the guardrail based on machine visual technology,and provides early warning according to the distance of the guardrail and the direction of the driving lane and the lane line.The method installs a binocular camera in the left front of the cleaning vehicle,collects depth image information of the guardrail,and obtains a distance value bet ween the guardrail and the binocular camera through the depth image;installs a monocular camera in front of the vehicle to obtain guardrail information in front of the vehicle.And the lane line information,the angle between the vehicle's driving route a nd the lane line can be calculated by an algorithm.The warning system warns the driver according to the distance information of the guardrail and the angle between the driving room and the lane of the vehicle,prompting the driver to correct the position of the cleaning vehicle in time to ensure the safe operation of the guardrail cleaning operation.The main contents of this paper are as follows:1)By consulting relevant literature,we can understand the current principle of ranging such as millimeter wave radar,laser radar and binocular stereo vision,and their advantages and disadvantages.According to the accuracy requirements and cost requirements of the early warning system,choose binocular stereo.Visually measure the fence.The principle of binocular stereo vision ranging and image filtering algorithm are introduced in detail,and the error analysis of binocular stereo vision ranging results is carried out.2)By using the Watershed Algorithm to detect the guardrail in front of the cleaning vehicle,it is possible to detect whether there is a guardrail on the side of the vehicle.When there is a guardrail,it can be determined whether the guardrail enters the blind spot of the binocular vision.Through the image edge detection of the improved Canny algorithm,the lane line is detected by using the Hough transform,and the angle between the vehicle traveling direction and the lane line is obtained by an algorithm.3)The design warning program realizes the early warning based on the fence distance,the information of the vehicle and the lane line angle,and completes the design visualization interface with Qt software.The guardrail distance value measured by the binocular stereo vision,the guardrail detection information and the lane line information are transmitted to the early warning program through the Socket communication method under the TCP protocol,and the early warning function is realized.4)The algorithm is transplanted on the embedded platform Jetson TX1,and the accuracy and feasibility of the real vehicle test and early warning system are carried out.The actual vehicle test results show that the vision-based guardrail cleaning vehicle safety auxiliary early warning system can measure the distance of the guardrail in real time and accurately,and according to the guardrail distance warning prompt.It can effectively improve the work efficiency and safety of the cleaning vehicle,and is of great significance to the intelligent development of sanitation engineering vehicles.