Research On Unmanned Vehicles Obstacle Ahead Detection Method

The unmanned vehicle can independently travel without manipulator in the road or field environment. With the continuous development of detection technology and improvement of human living standards, the unmanned vehicle technology has been paid more and more attention to.Obstructions detection in the perception system is a key component of unmanned vehicle autonomous driving, which directly affects the safety and reliability of unmanned vehicles traveling independently. Unmanned vehicle obstructions detection methods in the road environment has been discussed deeply in this paper. Some of the results achieved as follows:Firstly, a real-time detection method of dynamic obstructions in the road environment is presented. Two types of sensors, laser radar and GPS in-car, are used to detect the dynamic obstructions. Laser radar is used to obtain two continuous frames data and GPS in-car is used to attain heading angle and location information of unmanned vehicle between the obtained two frames. Matching the two frames data to find the co-exist obstructions and calculating the moving orientation can give a forecast area. The experimental results verified effectiveness of the algorithm and demonstrate that the method is meaningful with a certain sense of unmanned vehicles in advance to avoid dynamic obstructions.Secondly, rapid joint calibration method of laser radar and camera have been studied. The camera is calibrated by using board calibration plane and then joining with the laser data. The joint calibration has a better experiment result,which is simple and practical and can achieve higher accuracy.Thirdly, the obstructions detection method based on the camera and laser radar in the road environment is presented. The laser data can provide precise information of obstructions distance but can not access the shape features while the camera data can provide the shape of the feature information but can not obtain its depth of information. This method first picks laser radar and camera data, then projects onto the image after preprocessing and classifying the laser data. Moreover, obtaining and processing the obstructions detection areas in the image using the distance of the laser data in order to get the obstructions’edge. Finally, calculating maximum the height and width of the obstructions. The experiment shows that the error of obstructions detection is small enough to use in practical.