Research And Implementation On Unmanned Vehicle Obstacle Detection Algorithm Based On FPGA Stereo Vision

The perception and understanding of the driving environment is the premise and foundation of path planning for unmanned vehicle.The obstacle detection method is one of the important research objects of the environment perception system,and it is also one of the major challenges in this field.The road scene disparity map is obtained through the binocular stereo vision system.It also can reduce the road environment.The method has the advantages of direct,abundant information,strong robustness.After the method result,the feasible region of the road surface is extracted and the ROI region is extracted by using the V-disparity map theory.This method is simple and effective.Unmanned vehicle is a movable platform,which requires that the system has the advantages of small size,low energy consumption.Binocular stereo matching algorithm has the characteristics of large amount of computation and complex matching process.General purpose computer is unable to meet the requirements of obstacle detection system because of its operation flow,structure and energy consumption.Therefore,it is important to study how to improve the operation speed of the stereo matching algorithm,the algorithm of V-disparity detection,and how to reduce the energy consumption.Based on the FPGA algorithm,the paper studies and realizes the stereo matching algorithm and obstacle detection algorithm for road surface environment.The main contents and contributions are as follows:Real time SGM algorithm based on FPGA platform is designed.In view of FPGA hardware platform,based on the original semi global matching algorithm,through the optimization of the structure of the algorithm,the algorithm structure is running in the pipeline state.It achieves highly real-time performance;Through simplify the structure of the algorithm,this paper makes the algorithm can be applied to the low power consumption of the FPGA platform.Using the fixed model of road surface and V-disparity to detect obstacle.Aiming at the road environment,the simply method is proposed.The near distance road surface can be consider as a horizontal plane.The traditional algorithm based on the V-disparity image to estimates the pavement model.Using fixed road model,the paper proposes simplified algorithm.Compared with the traditional algorithm,the algorithm is more efficient.It can speed up the algorithm running.The stereo vision algorithm IP core is designed by HLS.HLS is also known as the high level language synthesis or behavior synthesis.It can be directly synthesis the high level language into the RTL FPGA hardware description language.HLS is far more efficient than the traditional methods in the software development.In this paper,the HLS Vivado software is used to develop the algorithm,and the process and structure of the stereo matching algorithm is designed,and the algorithm is simulated and running on the platform.Through the above work,this paper makes a deep research on the obstacle detection algorithm based on FPGA,and proposes the corresponding real-time and efficient solution from the stereo matching algorithm,V-disparity detection algorithm and HLS development method.After the simulation,the platform verification and performance analysis,the obstacle detection algorithm designed in this paper can be applied to the mobile platform,and has a good prospect in its application field.