Road Detection And Obstacle Avoidance Technology Based On GNSS And Vision

With the wide popularity of vehicles, traffic congestion, traffic accident-prone, and even the environmental pollution have become a problem that people have to face. In order to ease traffic congestion, reduce accidents and improve the efficiency of road transportation, Intelligent Transportation Systems(ITS) came into being. Among them, the intelligent vehicle positioning technology, road detection technology, and obstacle avoidance path planning technology are several critical technologies to ITS. Existing studies on intelligent vehicle are based on the radar detectors or on the laser range finder. These sensors are efficient and high precision, but expensive. We often need to make a compromise choice between economic and precision.The cost of vision sensor is low. With the development of image processing and machine vision, binocular vision sensor is widely used in various fields. This thesis will complete the intelligent vehicle positioning, road detection and avoidance path planning with the binocular vision sensor and GNSS sensor.The main contents and innovation points of this thesis are as follows:(1) Establish the binocular vision system. Take a checkerboard panel as a calibration-object to finish the calibration of the left and right cameras and solve the station P’s coordinates in the camera coordinate system and horizontal coordinate system, namely the positioning of station P relative to the intelligent vehicle. Finally, with the Bursa-Wolf model, the intelligent vehicle absolute positioning is finished based on the known station P’s absolute coordinates in WGS-84 coordinate system.(2) After the completion of the intelligent vehicle absolute positioning based on binocular vision sensor, it solves the fused coordinates of the intelligent vehicle with the vision positioning result and GPS positioning result through the use of kalman filter. The experiments show that higher accuracy positioning result is obtained after fusion.(3) An adaptive unstructured road detection algorithm based on HSI color space is proposed. The algorithm is for rural unstructured road. The road image is converted from RGB color space to HSI color space. And with the use of analysis and conclusions: 1, the difference of hue(H) between the road region and the non-road region. 2, the difference of saturation(S) between the shaded region of road and non-shaded region, the algorithm adaptively adjusts the ratio of saturation in HSI color model image based on priori information. As a result, the algorithm excludes the impact of the shaded area and extracts the full road area better when detecting the road.(4) An improved artificial potential field obstacle avoidance path algorithm under dynamic scene is proposed. Compared to the traditional artificial potential field, it is applied to the outdoor intelligent vehicle obstacle-avoidance-path-planning. To simulate real-world scene, the road constraints is added, the intelligent vehicle sensing range is limited to 150 °fan-shaped area and the most common local minima problem is solved effectively.