环境感知MEMS激光雷达固态与全景成像关键问题研究

With the development of intelligent mobile robot and autonomous vehicle technology, it is urgent to solve the problem of the solid state of the existing environmental perception lidar, which is lidar can be truly integrated into the vehicle through removing the heavy mechanical rotating parts. Mechanical rotating lidar plays an important role in the autonomous driving concept car, but it only could be solid by the sacrifice of the detection distance. Optical phased array laser radar is an ideal solid state technology route, but enslaved to the diffraction effect, the detection distance can only be more than ten meters. This project proposes a solid and panoramic imaging technology route, which has advantages of both mechanical and optical phased arrays: MEMS micro mirror is placed inside the omnidirectional optical system to solve the problem of solid state, which is also known as mixed solid; MEMS micro mirror spiral omnidirectional scanning, which has rotating 360 degrees horizontal field of view, and the number of turns to achieve more than 20 degrees vertical field of view. Problems to be solved: dynamic analysis and tracking control method of electrostatically actuated MEMS micro mirror driven by external parameters; helix high frame rate and high filling ratio of omnidirectional scanning technology of MEMS micro mirror; high sensitivity detection technique for speckle suppression; digital processing method of pulse echo centroid timing and time interpolation measurement; optimizing the parameters of each subsystem of the MEMS lidar. Finally, the key technical issues to be solved will be demonstrated.

随着智能移动机器人自主驾驶车技术的发展，迫切需要解决现有环境感知激光雷达的固态化难题，即去除笨重的机械旋转部件，将其真正集成到整车当中。机械转式激光雷达在自主驾驶概念车中发挥了重要作用，但要想固态化，只能牺牲探测距离。光学相控阵激光雷达曾是一种理想的固态化技术路线，但受制于衍射效应，探测距离目前也只能做到十几米。本项目提出的固态化与全景成像技术路线，兼有机械转式和光学相控阵的优点：MEMS微镜置于全景光学系统内部，解决固态化难题，亦称为混合固态；MEMS微镜螺旋线全景扫描，旋转实现360度水平视场，旋转圈数实现20度以上垂直视场。拟解决：外参数激励静电驱动MEMS微镜动力学分析与追踪控制方法；MEMS微镜螺旋线高帧频高填充比全景扫描技术；抑制散斑效应的高灵敏度探测技术；脉冲回波形心定时与时间内插测时数字处理方法；优化环境感知MEMS激光雷达各子系统参数。最终演示解决的关键科技问题。

本项目以人工智能系统的环境感知为需求背景，开展了激光雷达固态化和全景成像等科学问题研究，解决了外参数激励静电驱动MEMS微镜动力学分析与追踪控制、MEMS高帧频高填充比大角度扫描、抑制散斑效应的高灵敏度激光探测、脉冲回波形心定时与测时的激光测距信号处理等若干关键科学与技术问题，研制出大视场MEMS环境感知激光雷达样机。.1.主要研究内容.(1)外参数激励静电驱动MEMS微镜动力学分析与追踪控制.建立静电驱动MEMS扭转动力学模型，在外参数指令信号激励下，MEMS扫描镜扭转状态经历阻尼运动、周期、倍周期和混沌等动力学过程，其中周期运动存在着一个较大的参数控制范围。根据 Lyapunov稳定性理论，设计了一种连续混沌系统追踪控制器，受控的MEMS扫描镜扭转状态收敛任意期望的指令信号。.(2)MEMS高帧频高填充比大角度扫描.提出上下双层结构全景扫描光学系设计方法， MEMS螺旋轨迹指令控制实现了水平全景扫描，上下双层结构实现了大角度垂直扫描；提出MEMS微镜块扫描组合设计方法，实现了大角度高质量线性准静态扫描。.(3)抑制散斑效应的高灵敏度超宽带激光探测.建立大尺度留数散斑理论模型，逼近真实场景地仿真出激光束传输和反射特性；小尺度接收光学系统组合设计，抑制近场散斑效应，提高探测灵敏度；阵列式探测器芯片靶面设计方法，构建较大面积探测器芯片，大视场探测，小面积芯片单元超宽带，纳秒量级脉冲波形保真，高精度定时与测时信号处理。.(4)脉冲回波形心定时与测时的激光测距信号处理.提出了一种自适应波形心定时算法，结合机器学习选择斜率和寻峰，计算出波形的质心，在信噪比10的条件下，定时精度高达83ps；采用光触发和时间数字转化测时电路，测时精度高达130ps，测距精度约为2cm。.2.重要成果.该项目已完成计划任务书中所规定的全部研究内容，发表期刊论文28篇(国外期刊论文26篇，国内期刊论文2篇，其中包括国际顶刊论文2篇)、授权国家发明专利4项、出版译著2本。毕业答辩博士生1名、硕士生4名。部分研究成果已实现转化。研制的大视场MEMS激光雷达样机，探测距离150m、测距精度5cm、视场50度*25度、角分辨率0.1度*0.5度、帧频15Hz。

内容获取失败，请点击重试