激光与光电子学进展, 2023, 60 (7): 0726002, 网络出版: 2023-04-13
偏振光辅助的视觉惯性导航系统
Visual Inertial Navigation System Aided by Polarized Light
组合导航 偏振定向传感器 单目视觉 同步定位与建图 图像处理 integrated navigation polarization orientation sensor monocular vision simultaneous localization and mapping image processing
摘要
为了提高组合导航系统的可靠性与位姿估计的精度,把偏振定向传感器引入同步定位与建图过程,提出并设计一种新的偏振光辅助的视觉惯性组合导航系统。采集偏振定向传感器、单目视觉相机及微惯性测量单元的数据,对多传感器数据进行时间戳对齐和预处理后,利用最小二乘优化方法建立目标方程,通过求解非线性方程组获取最佳的运动估计。该系统根据天空偏振分布实现了方位角的可观性,并融合了多传感器数据。基于上述组合导航系统进行户外车载实验,实验结果表明:在2 km的长距离运行中,相比原始视觉惯性系统,偏振光辅助的视觉惯性导航系统的位置误差降低了16.7%,航向角精度提升了23.4%。偏振定向传感器的接入能够抑制惯性器件测量值的漂移,改善导航系统的位置精度和姿态角精度,可满足卫星信号受到干扰等环境下的位姿估计精度和可靠性要求。
Abstract
To improve the reliability of the integrated navigation system and the accuracy of pose estimation, a novel visual inertial integrated navigation system assisted by polarized light is proposed and constructed by introducing a polarization orientation sensor into the process of simultaneous localization and mapping. The data from the polarization orientation sensor, monocular vision camera, and micro-inertial measurement unit are collected. The target equation is established using the least square optimization method after multi-sensor data has been time-stamped aligned and preprocessed, and the best motion estimation is then produced by solving nonlinear equations. In the suggested system, the observability of azimuth is achieved based on the polarization distribution of the sky, and multi-sensor data are fused. Based on the above-integrated navigation system, an outdoor vehicle-mounted experiment is carried out. The experimental findings indicate that, in the long-distance operation of 2 km, the location inaccuracy of the polarized light-assisted visual-inertial navigation system is 16.7% lower and the heading angle accuracy is 23.4% greater than the estimated value of the original visual-inertial system. The polarization orientation sensor can reduce the drift of inertial devices, enhance the position accuracy and attitude angle accuracy of the navigation system, and meet the needs of the position and pose estimation accuracy and reliability under the interference of satellite signals.
褚金奎, 胡瀚珮, 万振华, 李金山. 偏振光辅助的视觉惯性导航系统[J]. 激光与光电子学进展, 2023, 60(7): 0726002. Jinkui Chu, Hanpei Hu, Zhenhua Wan, Jinshan Li. Visual Inertial Navigation System Aided by Polarized Light[J]. Laser & Optoelectronics Progress, 2023, 60(7): 0726002.