Design And Simulation Studies Of Vehicle Autopilot System Based On FPGA

With the advancement of technology, from the day the car was born, peoplenever stop automotive technology research. In recent years, research onautomotive technology began to develop intelligence and automation, smart carsand autopilot technology came into being. This paper is based on FPGAapplication technology and digital image processing technology, research thevehicle autopilot system, the entire system has two parts, lane automaticidentification system and autopilot systems. System applicating FPGAtechnology uses a top-down hierarchical design principle, completes systemsimulation tests.In the path identification section, the digital image processing technology isused in the path information of the vehicle collected for processing, innovativelyproposed a " double split iterative algorithm of OTSU ", and parallel thresholdselection algorithm is designed to complete the road image binarizationprocessing;" Fitting method of the curve line of least squares approximationbased on line approximation" is proposed innovativly to complete the lanerecognition system. FPGA hardware implementation of this section is simulationresults show that on the lane line identification has good real-time and accurateidentification accuracy.In the autopilot system, binocular Vision Technology and camera technologyare combined for the characteristics of lane information, on the principle ofbinocular vision the algorithm of target point location algorithm is improved toachieve the transformation of dimensional image coordinates to worldcoordinates; on the basis of the detailed analysis of the process during brakingand overtaking, the calculation model of vehicle safety distance between vehiclesis finally determined; in the system control section, according to thecharacteristics of vehicle control and an experienced pilot experience, the fuzzy rule and controlled conditions are designed, to achieve automatic control functionof the vehicle. In this section, because of the coordinate transformation, pathplanning and control algorithms is complex, computationally intensive, so thispart is implemented in use of the FPGA NIOS II soft-core.In the system simulation stage, we used the Modelsim software to completesimulation of each module separately, observe the waveform and determinewhether the desired effect. Finally, with the use of virtual reality technology wecreated a vehicle and driving environment model to simulate the real vehicle,system-level simulation.All stages of the simulation results show that the algorithm can achieve theprocessing speed of this real-time requirement and high recognition accuracy andcontrol, with good reliability and robustness...