Design And Implementation Of Hardware Platform For Multi-sensor Data Acquisition And Positioning Algorithm Acceleration Of Driverless

Improving the synchronization accuracy of multi-sensor data acquisition and the accuracy and calculation speed of map construction positioning algorithms is one of the key technologies of driverless systems.driverless systems have become a research hotspot in the high-tech industry,however,the synchronous collection and precise fusion of data from a variety of sensors and hardware acceleration of positioning algorithm for driverless vehicles in the current driverless systems is a relatively independent platform.The processing and analysis platform of multi-sensor data mostly relies on the cloud or terminal computer,so there are shortcomings such as high total cost,large size,high power consumption,and unsatisfactory vehicle-level applications.Aiming at the above problems,This paper designs a hardware platform for driverless multi-sensor data acquisition and algorithm acceleration based on a high-speed parallel Field Programmable Gate Array(FPGA)as the core and powerful computing power Jetson TX2.The hardware platform is researched from the aspects of low cost,multi-sensor data synchronously acquisition control and scheme of hardware accelerated platform for high precision positioning algorithm.The specific research contents of this paper are as follows:(1)An integrated solution for multi-sensor data synchronous acquisition and high-precision positioning algorithm acceleration platform is designed.This system uses FPGA as the core controller and the small industrial computer Jetson TX2 as the data processing unit.It adopts a variety of sensor integration solutions to fully perceive the environmental information,including dual-frequency Global Navigation Satellite System(GNSS),inertial measurement unit(IMU),Odograph(ODO),camera,Three-dimensional lidar and other sensors,which make up for the shortcomings that single sensor cannot get environmental information comprehensively.At the same time,the main controller FPGA can be used as a hardware accelerator for driverless high-precision positioning algorithms,and a high-speed DDR3 memory interface is designed to meet the storage requirements during the process of algorithm acceleration.(2)The design of multi-sensor data collection and high-precision positioning algorithm hardware acceleration platform schematic diagram,PCB board drawing and debugging work are completed.Designed the entire power module,the peripheral circuit of the main control FPGA,the peripheral interface circuit of Jetson TX2,the data acquisition and synchronization control circuit of various sensors,and the high-speed communication interface circuits such as USB,Gigabit Ethernet,PCIe and DDR3,mSATA,TF card storage circuits.Finally,the entire PCB board is debugged and verified to provide multi-sensor synchronous data for driverless cars,and can provide a hardware acceleration platform for driverless high-precision positioning algorithms.(3)high-precision synchronous acquisition of multi-sensor data is completed.This system uses Pulse Per Second(PPS)of Global Positioning System(GPS)and high-stability quartz crystal to establish a high-precision time reference:the high stability quartz crystal oscillator with short-term stability is used as the clock signal input of FPGA,and the FPGA with high-speed parallel operation simulates the clock operation,at the same time,the PPS second pulse and NEMA data output by GPS with long-term stability are used to refresh the entire system clock.By this means to achieve absolute unity in time.Taking the time reference as a link,different synchronization methods are adopted according to different characteristics of different sensors.Choose a lower cost BMI088 inertial measurement unit,and complete different sensor acquisition circuits according to the sensor transmission interface.(4)The debugging of high-speed serial bus CSI interface camera and DDR3 high-speed memory circuit are Completed.In order to make up for the shortcomings of CSI cameras that cannot transmit in a long distances,the deserializer hub scheme of DS90UB954-Q1 is designed,The long-distance transmission is realized by receiving serial sensor data through the FPD-Link ? interface.The design of DDR3 read and write control module based on XILINX MIG IP core is completed,and verified by simulation.The experimental results show that the platform can provide high-precision synchronization data including GPS,IMU,ODO,camera,three-dimensional lidar sensors for driverless systems,and the synchronization accuracy is controlled at the microsecond level.At the same time,this design can provide a hardware acceleration platform for location algorithms related to driverless vehicles.This design has the advantages of high integration,high synchronization accuracy,low cost,and wide range of applications.It can be applied not only to driverless systems,but also in UAV,positioning navigation,mobile measurement and other fileds.It has broad application prospects.