Real-time simulation of time-of-flight sensors
Real-time simulation of time-of-flight sensors
复制标题
DOI：
10.1016/j.simpat.2009.03.004
复制
发表时间：
2009-05-01
影响因子：
4.2
通讯作者：
Kolb, Andreas
中科院分区：
文献类型：
--
作者：
Keller, Maik;Kolb, Andreas
今天的飞行时间(TOF)传感器通过实时估计从发射到接收主动光之间经过的时间来测量全范围距离信息。这类传感器价格低廉、体积小、性能高，特别适合实时应用，例如在汽车、机器人、3D成像和可视化领域。这类传感器的模拟是硬件设计和应用程序开发的重要组成部分。因此，仿真数据必须能够反映传感器的主要特性。针对一种特殊类型的飞行时间传感器--光子混合器件(PMD)传感器，提出了一种基于物理激励的仿真方法。动态运动模糊和分辨率伪影(如飞行像素)以及典型的偏差误差是真实系统的突出影响。当不均匀深度的区域被单个PMD像素覆盖时，会出现漂移像素，而偏差误差是基于光信号的非谐波特性。这些人工产物的建模对于真实的模拟方法是至关重要的。利用现代图形处理器(GPU)的可编程性，提出了一种硬件加速方案，并对实际PMD器件和仿真数据进行了详细的比较。这样，为应用和进一步的数据处理提供了交互的模拟反馈。仿真是实时进行的，因此也可以实时访问所有所需的控制机制。(C)2009爱思唯尔B.V.保留所有权利。
Today's time-of-flight (TOF) sensors measure full-range distance information by estimating the elapsed time between emission and receiving of active light in real-time. Such sensors are inexpensive, compact, and they have a high performance, which especially fits real-time applications, e.g. in the fields of automotive, robotics, 3D imaging, and visualization. The simulation of such sensors is an essential building block for hardware design and application development. Therefore, the simulation data must capture the major sensor characteristics.This paper introduces a simulation approach, which is motivated by physics, for the Photonic Mixing Device (PMD) sensor which is a specific type of time-of-flight sensor. Dynamic motion blurring and resolution artifacts such as flying pixels as well as the typical deviation error are prominent effects of real world systems. Flying pixels arise when an area of inhomogeneous depth is covered by a single PMD-pixel whereas the deviation error is based on the anharmonic properties of the optical signal. The modeling of these artifacts is essential for an authentic simulation approach. We present a detailed comparison between a real PMD-device and the simulation data regarding the sensor characteristics.The proposed algorithms are implemented in a hardware accelerated solution which makes use of the programmability of modern Graphics Processing Units (GPUs). This way, an interactive simulation feedback is provided for applications and further data processing. The simulation takes place in real-time and thus all required control mechanisms are accessible in real-time, too. (C) 2009 Elsevier B.V. All rights reserved.