Characterization of a RS-LiDAR for 3D Perception
Zhe Wang, Yang Liu, Qinghai Liao, Haoyang Ye, Ming Liu and Lujia Wang
Abstract— High precision 3D LiDARs are still expensive and
hard to acquire. This paper presents the characteristics of RS-
LiDAR, a model of low-cost LiDAR with sufficient supplies,
in comparison with VLP-16. The paper also provides a set of
evaluations to analyze the characterizations and performances
of LiDARs sensors. This work analyzes multiple properties,
such as drift effects, distance effects, color effects and sensor
orientation effects, in the context of 3D perception. By compar-
ing with Velodyne LiDAR, we found RS-LiDAR as a cheaper
and acquirable substitute of VLP-16 with similar efficiency.
I. INTRODUCTION
A. Motivation
Along with the rapid development of autonomous driving,
multi-beam LiDAR has become one of the most important
sensors on autonomous cars. Light Detection and Ranging,
known as LiDAR, is a system using lasers to mainly detect
the geometrical properties such as location, shape, and ve-
locity. It beams lasers to the target objects and receives the
signals reflected by those objects. By comparing the phase
difference between the received signals with the sent ones,
it reveals information about the target objects, for example,
the distance, reflectivity, etc. By adopting further algorithms,
the position, altitude, velocity, pose, and shapecould also be
obtained [4], [5]. Generally, LiDAR is capable of detecting
targets with a precision of several centimeters.
LiDAR on vehicles is the critical sensor that serves for
mapping and localization, for which multiple products have
been developed. Velodyne LiDAR is widely considered to
be the most popular LiDAR company who has released
several LiDAR products such as HDL-64E, HDL-32E and
VLP-16 for mapping and 3D perception purposes [6]. VLP-
16 has generated significant interest in the surveying and
mapping industry because of its compact size, low power
requirements, and high performance. However, Velodyne
LiDAR are comparably expensive and buyers should wait
for at least 6 months to get the sensor. At present, plenty of
low-cost LiDAR products have come out and two of the most
famous ones are RS-LiDAR from Robosense and PANDAR
40 from HESAI.
For RS-LiDAR is also very popular in autonomous driving
companies like TuSimple
1
, and RoadStar
2
, we evaluate its
characteristics and justify wether it is feasible as replacement
Zhe Wang, Yang Liu, Qinghai Liao, Haoyang Ye and Ming Liu are with
Robotics and Multi-Perception Lab (RAM-LAB), Robotics Institute, The
Hong Kong University of Science and Technology
Lujia Wang is with Cloud Computing Lab of Shenzhen Institutes of
Advanced Technology, Chinese Academy of Sciences
1
http://www.tusimple.com
2
http://roadstar.ai
Fig. 1: RS-LiDAR from Robosense
(Retail price: about $7000)
of the VLP-16. As the same as VLP-16, RS-LiDAR is a 16-
channel real-time 3D LiDAR with a similar dimensions and
weights. The RS-LiDAR is shown in Fig.1 is designed to
be used on autonomous cars, robots, and UAVs. Usually,
these applications require high accuracy and proper size and
weight. With knowledge of the parameters in the official
handbook, we consider it is still necessary to have a system-
atic evaluation of its accuracy, repeatability, and stability.It is
worth to mention that there are some low-cost 2D LiDARs
such as RpLiDar A2 and A1 from SlamTec
3
, which can be
assembled with rotational parts and work as 3D sensor. As
the scanning frequency is much lower comparatively, we only
consider the 3D LiDARs in this paper.
B. Related Work
For laser scanner evaluation, many papers had been pub-
lished with different experiments for different sensors. Kneip
proposed a characterization and extended some specialized
tests with a subsequent calibration model of a 2D LiDAR,
URG-04LX [4]. Glennie put forward a calibration and sta-
bility analysis of the VLP-16 laser scanner [5]. Ye and
Borenstein presented a characterization study of the Sick
LMS 200 laser scanner [11]. Stone reviewed the basic
physics and implementation of various LADAR technologies,
describing the problems associated with available ’off-the-
shelf’ LADAR systems and summarizing worldwide state-
of-the-art research. He also elaborated on general trends
in advanced LADAR sensor research and their likely im-
pact on manufacturing, autonomous vehicle mobility and on
construction automation [12]. Kawata introduced a method
to develop an ultra-small lightweight optical range sensor
system [13] and Ueda proposed an accurate range data
mapping system with sensor motion [14]. According to their
tests and analyses, we detected the commonly concerned
issues with RS-LiDAR and VLP-16.
3
https://www.slamtec.com
arXiv:1709.07641v1 [cs.RO] 22 Sep 2017
评论0