On Solving Mirror Reflection in LIDAR Sensing PDF20160222-963-15FIGA2

Title	On Solving Mirror Reflection in LIDAR Sensing
Author	Chieh-Chih Wang
Pages	11
File Size	1024 KB
File Type	PDF20160222-963-15FIGA2
Total Downloads	334
Total Views	487

Summary

IEEE/ASME TRANSACTIONS ON MECHATRONICS, VOL. 16, NO. 2, APRIL 2011 255 On Solving Mirror Reflection in LIDAR Sensing Shao-Wen Yang, Student Member, IEEE, and Chieh-Chih Wang, Member, IEEE Abstract—This paper presents a characterization of sensing fail- LIDARs are appropriate for high-precision appli...

Description

IEEE/ASME TRANSACTIONS ON MECHATRONICS, VOL. 16, NO. 2, APRIL 2011 255 On Solving Mirror Refeetion in LIDAR Sensing Shao-Wen Yang, Student Member, IEEE, and Chieh-Chih Wang, Member, IEEE Abstract—This paper presents a characterization of sensing fail- ures of light detection and ranging (LIDAR) given the presence of a mirror, which are quite common in our daily lives. Although LIDARs play an important role in the feld of robotics, previous research has addressed little regarding the challenges in optical sensing such as mirror refections. As light can be refected of a mirror and penetrate a window, mobile robots equipped with LIDARs only may not be capable of dealing with real environ- ments. It is straightforward to deal with mirrors and windows by fusing sensors of heterogeneous characteristics. However, in- distinguishability between mirror images and true objects makes the map inconsistent with the true environment, even for a robot with heterogeneous sensors. We propose a Bayesian framework to detect and track mirrors using only LIDAR information. Mir- rors are detected by utilizing the property of mirror symmetry. Spatiotemporal information is integrated using a Bayesian flter. The proposed approach can be seamlessly integrated into the occu- pancy grid map representation and the mobile robot localization framework, and has been demonstrated using real data from a LIDAR. Mirrors, as potential obstacles, are successfully detected and tracked. Index Terms—Mobile robots, optical scanners, range sensing, sensor fusion. I. INTRODUCTION SIMULTANEOUS loealization and mapping (SLAM) is the proeess by whieh a mobile robot ean build a map of the environment and, at the same time, use this map to eompute its loeation. As the SLAM problem has attraeted immense atten- tion in the mobile roboties literature, a large variety of sensors have been used for SLAM, sueh as sonar, light deteetion and ranging (LIDAR), IR, monoeular vision, stereo vision, and GPS. The past deeade has seen rapid progress in solving the SLAM problem [1], [2], and LIDARs are at the eore of most state- of-the-art robot systems, sueh as Boss [3] and Stanley [4], and the autonomous vehieles in the Defense Advaneed Researeh Projeets Ageney (DARPA) Urban Challenge and Grand Chal- lenge. Beeause of their narrow beamwidth and fast time of fight, Manuseript reeeived February 2, 2009; revised June 6, 2009, Oetober 2, 2009, and Deeember 7, 2009; aeeepted Deeember 25, 2009. Date of publieation February 8, 2010; date of eurrent version January 19, 2011. Reeommended by Teehnieal Editor C. A Kitts. This work was supported in part by the Taiwan National Seienee Couneil under Grant 96-2628-E-002-251-MY3, Grant 96- 2218-E-002-035, Grant 97-2218-E-002-017, and Grant 98-2218-E-002-006, in part by the Exeellent Researeh Projeets of the National Taiwan University, in part by Taiwan Miero-Star International, in part by Compal Communieations, Ine., and in part by Intel. S.-W. Yang is with the Department of Computer Seienee and Informa- tion Engineering, National Taiwan University, Taipei 10617, Taiwan (e-mail: [email protected]). C.-C. Wang is with the Department of Computer Seienee and Information Engineering and the Graduate Institute of Networking and Multimedia, National Taiwan University, Taipei 10617, Taiwan (e-mail: [email protected]). Color versions of one or more of the fgures in this paper are available online at http://ieeexplore.ieee.org. Digital Objeet Identifer 10.1109/TMECH.2010.2040113 LIDARs are appropriate for high-preeision applieations in the feld of roboties. A LIDAR estimates the distanee to a surfaee by measuring the round-trip time of fight of an emitted pulse of light. Only a fraetion of the photons emitted by the LIDAR are reeeived baek through the sensor's opties, with this amount being a strong funetion of the refeetivity of the objeet being imaged. Table I summarizes how surfaee properties afeet the amount of light refeeted, absorbed, and transmitted. White surfaees re- feet a large fraetion of light, while blaek surfaees refeet only a small amount. Transparent objeets sueh as glasses often refraet the light, and a LIDAR measurement of sueh a surfaee typieally results in the range information for the objeet behind the trans- parent surfaee. In addition, the mirror-like refeetion of light, in whieh light from a single ineoming direetion is refeeted into a single outgoing direetion, is ealled speeular refeetion or regular refeetion. This is in eontrast to difuse refeetion, where light bounees of in a number of angles due to the irregularity of a surfaee. Mirrors are very fat surfaees and refeet nearly all in- eident light sueh that the angles of ineidenee and refeetion are equal. In geometry, the mirror image of an objeet is the virtual image formed by refeetion in a plane mirror. The mirror image that is formed appears to be behind the mirror and is of the same size as the real objeet illuminated by the LIDAR via the mirror. Figs. 1 and 2 illustrate the eireumstanees of mirror refeetion and glass transpareney. As a result, deteetion of mirrors and windows ean be problematie in laser sensing [5] and [6]. To our best knowledge, the solution to the problem of mirror refeetion has not been addressed yet. As LIDARs have beeome the major pereeptual sensors, mirrors and windows ean pose a real danger to robots with limited pereeptual eapability. In this paper, the problem of mirror refeetion is addressed. The main eontribution of this study is to provide a solution to deteet and traek mirrors using only LIDAR information. The mirror deteetor utilizes the geometrie property of mirror sym- metry to generate hypothetieal mirror loeations. An identifed mirror loeation is represented using a line model with endpoints. The mirror traeker is then used to integrate the potential mirror loeations temporally using a Bayesian flter. The spatiotemporal information is aeeumulated and used to provide reliable seene understanding. A Bayesian framework is also introdueed to the mobile robot mapping and loealization proeess so that the mir- ror images ean be eliminated. The proposed approaeh has been demonstrated using real data from the experimental platform equipped with a SICK LMS 291 LIDAR, as shown in Fig. 1. The performanee of the proposed approaeh has also been evalu- ated using real data. The ground truth is obtained using another LIDAR that ean observe the aetual boundary of a mirror. The ample experimental results demonstrate the feasibility and ef- feetiveness of our approaeh. 1083-4435/$26.00 © 2010 IEEE...