【作者】 刘洋；

【导师】 何晶；

【作者基本信息】 湖南大学 ， 信息与通信工程， 2023， 硕士

【摘要】 在未来万物互联的应用中,高效可靠的定位方案对于工厂机器人的自行导航、轨迹跟踪、无人车辆的自动驾驶、辅助驾驶等应用至关重要。当前全球定位系统(global positioning system,GPS)应用广泛,但由于链路阻塞和多径效应,使其在城市峡谷,隧道内或地下等场景中受到了严重的挑战。目前,基于可见光通信(visible light communication,VLC)的定位方法受到了研究人员的青睐,即可见光定位(visible light positioning,VLP)技术,其绿色低碳、定位精度高且无需额外的设备即可部署,可以大大降低实施成本,被认为是提供移动终端定位服务的有效解决方案之一。然而,基于VLP的移动终端定位方案在实际应用中仍存在着很多的问题,本文针对基于VLP的移动终端定位方案中存在的定位中断问题以及移动定位干扰问题进行了研究,进行了如下工作:1、提出了一种基于VLP的多传感器融合车辆定位方法。针对VLP系统中,发光二极管(light-emitting diodes,LEDs)稀疏分布引起的中断问题和定位算法的时间成本问题,提出了 一种粒子滤波(particle filtering,PF)辅助的基于单LED-VLP(single-LED VLP,SL-VLP)的惯性融合定位方法,提高了 VLP在稀疏LED场景中的鲁棒性。此外,研究了不同中断率和速度下的时间成本和定位精度。实验结果表明,采用所提车辆定位方法,在SL-VLP中断率为0%、5.5%、11%和22%时,平均定位误差分别为0.09 m、0.11 m、0.15 m和0.18 m。2、提出了一种移动环境下使用LED阵列灯的VLP方法。针对实际移动场景下,运动中的相机捕获的图像存在环境光的干扰,难以准确提取感兴趣信号区域,导致信号失真的问题,提出了一种基于LED阵列灯的感兴趣信号区域提取方法来提高可见光信号源的定位准确性,使用经过训练的卷积神经网络模型来减轻移动车辆捕获可见光信号源时存在的严重失真。实验结果表明,在车辆移动的情况下,所提出的VLP方法可以在不同速度下实现厘米级的定位精度。更多还原

【Abstract】 In future intelligent transportation systems,efficient and reliable vehicle localization is critical for automated driving,assisted driving,automated parking,and trajectory tracking services.Although the Global Positioning System(GPS)is widely used,it is severely challenged in urban canyons,tunnels,underground areas,and other scenarios due to link blockages and multipath effects.Recently a visible light communication(VLC)based positioning solution has been favored by researchers,namely visible light positioning(VLP)technology,which is green and low-carbon,has high positioning accuracy and can be deployed without additional equipment,which can greatly reduce the implementation cost.It is considered as one of the effective solutions to provide mobile terminal positioning services.However,VLP-based vehicle localization solutions still face many challenges in practical applications.In the paper,localization interruption and mobile localization interference issues in VLP-based vehicle localization solutions are studied and conduct the following work:1.A VLP-based multi-sensor fusion vehicle localization method is proposed.A particle filtering(PF)-assisted single-LED VLP(SL-VLP)based inertial fusion localization method is proposed to address the interruption problem caused by the sparse distribution of light-emitting diodes(LEDs)and the time cost of localization algorithms in VLP systems.)inertial fusion localization method,which improves the robustness of VLP in sparse LED scenes.In addition,the time cost and localization accuracy under different interruption rates and speeds are investigated.The experimental results show that the average localization errors are 0.09 m,0.11m,0.15 m and 0.18 m for SL-VLP interruption rates of 0%,5.5%,11%and 22%,respectively,using the proposed vehicle localization method.2.A VLP method using LED array lights in a moving environment is proposed.For the actual moving scenes,the images captured by the camera in motion have the interference of ambient light,which makes it difficult to extract the signal region of interest accurately and leads to signal distortion.An LED array light-based signal region of interest extraction method is proposed to improve the localization accuracy of visible light signal sources,using a trained convolutional neural network model to mitigate the severe distortion that exists when moving vehicles capture visible light signal sources.Experimental results show that the proposed VLP method can achieve centimeter-level localization accuracy at different speeds in the presence of moving vehicles.更多还原