The optical mouse for indoor mobile robot odometry measurement

https://doi.org/10.1016/j.sna.2005.09.015Get rights and content

Abstract

In this work the optical mouse is proposed for indoor odometry measurement in mobile robot applications. The optical mouse is a very low-cost sensor and has the advantage that the measured displacement is independent from the kinematics of the robot because the optical sensor uses external natural microscopic ground landmarks to obtain the effective relative displacement. In this work the sensor is calibrated and evaluated for odometry measurement. It was found that the original conception inside a visual feedback loop precludes its use as an isolated displacement sensor although its impressive speed and parts, with a CMOS camera and a digital signal processor embedded on the same chip, suggests that an improved design can be a good alternative for accurate mobile robot odometry measurement.

Introduction

Since its invention in 1964, the mouse has suffered from the same problems as the mobile robots. In a mobile robot, the displacement induced by mechanical devices uses to be measured to estimate its real displacement. In a mouse, the hand displacement has to be measured and converted into a cursor movement. The original mouse only had two wheels attached to encoders to measure horizontal and vertical displacement whereas the simplest and effective way to measure the displacement and trajectory of a mobile robot is to use an encoder placed in the drive motors [1], [2], [3], [4], in the active wheels or even in additional passive wheels [5]. However, encoder measurements suffer form problems of accurate wheel diameter estimation, uncertainty in the determination of the contact point with the floor, dirt accumulation and wheel wear and slippage.

For computer applications, these problems have little effect because the mouse it is included in a visual feedback loop where the user pays attention to the cursor in the monitor screen rather than to the mouse itself. This means that the relative trajectory of the mouse and cursor trajectory can differ; the visual feedback compensates possible errors in the measurements because only the cursor trajectory is important. The real mouse trajectory is only considered for ergonomics because reducing total mouse displacement reduces wrist stress and increases comfort. For this reason, the resolution provided by the mouse has been incremented from the initial 0.158 mm (160 counts per inch or cpi) to 0.0158 mm (1600 cpi) provided by the newest models.

In mobile robot applications, these problems play an important role because the goal is to measure the robot displacement with the maximum accuracy. To solve these problems additional sensors, such as ultrasonic [6], laser measurement system [7] and vision [8], [9] are widely used to extract and compare additional relative scenario information to correct the estimated robot position.

Vision researchers have placed on-board cameras in practically all the possible mobile robot locations with almost all possible orientations to obtain visual odometry measurements [9]. Probably the worst idea is to point the camera and all the computational resources of a dedicated vision system at the ground. However, in 1999 Agilent Technologies started to release an optical mouse with a small CMOS camera and a digital signal processor (DSP) embedded in the same chip. Using this sensor no additional computational resources are needed for a measurement of the displacement. The camera is pointed to the ground and takes snapshots while the DSP compares the snapshots to measure the relative camera (mouse) displacement over the ground. This innovative sensor design is contact free (is the mouse structure that contacts with the ground), does not suffer from wear and is immune to dirt because is flipped. In addition, the standard CMOS technology enables high chip integration [10] and low-cost production and thus a full replacement for the traditional mechanical mouse.

The mouse can be used as a displacement sensor [11], [12], [13]. The optical mouse has been proposed and tested as a two-dimensional displacement sensor [14] with three main conclusions: the optical mouse is unable to register displacements over transparent and reflective objects because there are no shadows under the mouse, there is a limitation of 1.25 mm in the maximum operative distance between the object and the optical sensor, and there are low levels of error and high degrees of linearity in the measurements. In [14] the influence of height on the measurements is evaluated in terms of linearity whereas its influence on resolution is not considered. In [14] the maximum displacement was limited to 1 mm, good enough for applications as vibratory motion sensing [15] but for odometry measurement we have extended the range to 50 mm. Recently, the optical mouse has been proposed specifically for odometry measurement in a mobile robot application [16] using two redundant sensors placed at different locations and with different orientations to compensate the error in the estimated displacement. In this work we identify the causes of these errors.

This paper investigates the problems originated by the use of the optical mouse as a mobile robot odometry measurement system and is organized as follows. Section 2 is dedicated to the working principle of the optical mouse; Section 3 shows the calibration of the optical mouse as an odometric sensor and the results are discussed in Section 4. Finally the work ends with some conclusions.

Section snippets

Optical mouse

The optical mouse is based on a main optical sensor with four main parts (Fig. 1): illumination device, illumination lens, camera and camera lens.

Illumination is an important problem for a vision system. A light emitting diode (LED) with wavelength peak from 639 to 875 nm in combination with a plastic lens and mirrors is used to illuminate the surface under the sensor. The chip contains a CMOS camera as the image acquisition system and a proprietary DSP for image processing and external

Optical mouse calibration

The following results were obtained using a standard optical mouse with a declared resolution of 0.0635 mm (15.74 mouse pulses per millimeter or 400 cpi). The optical sensor of the mouse is the HDNS-2000 type [19]. The mouse displacements were measured with a precision of ±0.05 mm and the maximum speed was limited to 1.5 mm/s to minimize the errors at the end of the displacements. The pulses generated by the mouse during the displacement were obtained using a MATLAB function [20] with all software

Discussion

We have found that the optical mouse is an unbiased estimator. The relative error obtained by applying the calibration equation to the distance measured with one sensor increases as the distance increases. The maximum relative error is below 0.8 mm in a 50 mm range (RMS error = 0.2 mm). However, averaging over 10 optical sensors the relative error is below 0.2 mm (RMS error = 0.02 mm) and does not depend on the distance. Therefore, the best solution for accurate distance measurements is to use an array

Conclusions

In this work, the use of the optical mouse for indoor mobile robot odometry measurement was studied. It was found that the linearity of the optical mouse is very good with a coefficient of determination of R2 = 0.9998 although the standard deviation of the measurements increases with the distance. However, in practice this deviation can be compensated by averaging the measurements of an array of sensors because the optical mouse is an unbiased estimator.

It was found that a small height variation

References (20)

  • E.J. Nijhof

    On-line trajectory modifications of planar, goal-directed arm movements

    Hum. Movement Sci.

    (2003)
  • T.W. Ng

    The optical mouse as a two-dimensional displacement sensor

    Sens. Actuat. A

    (2003)
  • T.W. Ng et al.

    The optical mouse for vibratory motion sensing

    Sens. Actuat. A

    (2004)
  • J. Borenstein

    Experimental results from internal odometry error correction with the OmniMate mobile robot

    IEEE Trans. Robotic Autom.

    (1998)
  • C. Tarin et al.

    Odometry error correction by sensor fusion for autonomous mobile robot navigation

  • A. Martinelli

    The odometry error of a mobile robot with a synchronous drive system

    IEEE Trans. Robotic Autom.

    (2002)
  • J. Palacin et al.

    Building a mobile robot for a floor-cleaning operation in domestic environments

    IEEE Trans. on Instrum. Measur.

    (2004)
  • J. Borenstein

    The CLAPPER: a dual-drive mobile robot with internal correction of dead-reckoning errors

  • N. Achour et al.

    Building an environment map using a sweeping system based on a single ultrasonic sensor

  • Z. Xu et al.

    Map building for indoor environment with laser range scanner

There are more references available in the full text version of this article.

Cited by (97)

  • Implementation and calibration of an odometry system for mobile robots, based on optical computer mouse sensors

    2020, Sensors and Actuators, A: Physical
    Citation Excerpt :

    Opposite to random errors that can compensate each other as the consequence of averaging, systematic errors will accumulate with each incremental step. The sensitivity of an optical computer mouse sensor tends to be inconsistent and depends on the distance between the sensor aperture and the floor surface and on properties of the sensor's motion trajectory such as velocity and curvature [7,10–12]. To remedy these limitations, modifications have been proposed to the sensor optics to render it afocal, effectively mitigating the height dependency of the sensitivity [13].

  • Preventing and correcting mistakes in lifelong mapping

    2021, 2021 10th European Conference on Mobile Robots, ECMR 2021 - Proceedings
View all citing articles on Scopus
View full text