Displacement Sensor using Laser Triangulation in Highway Pavement Inspection

Transcription

1 Displacement Sensor using Laser Triangulation in Highway Pavement Inspection Yinan Xing, Richard C. Liu ECE Department, University of Houston Houston, TX, 77204, USA ABSTRACT Traditional techniques like sand patch method and circular meter provide the reference value for pavement texture inspection. However, these traditional methods can not effectively fulfill the requirement of the highway pavement texture inspection. The non-contact displacement sensor that is able to provide more accurate and reliable results is desirable. The displacement sensor discussed in this paper is specifically designed for highway pavement inspection purpose. This paper mainly discusses the methods adopted by the sensor for eliminating the interference of ambient light and minimizing the error caused by the variation of the target characteristics. In this paper some simulation results and the experiment data were presented. Keywords: Displacement detection, Laser Triangulation, Position sensitive detector, Highway pavement, PID control 1. INTRODUCTION Pavement is the durable surface material intended to sustain traffic. Well designed and constructed pavements can provide safe, durable surfaces with low-noise characteristics [1]. Pavement surface texture influences many aspects of tire-pavement interaction, including friction, tire-pavement, noise, rolling resistance and tire wear [2]. Traditional technique, for example, sand patch method is used to provide the reference value, mean profile depth (MPD) [4] for the highway pavement inspection. However, this method is operator dependent, labor intensive and can not satisfy the accuracy and efficiency requirements. Therefore, a high-speed noncontact device that is able to measure the target movement is desirable for the highway pavement inspection. Laser technique has the characteristics of high accuracy, high resolution and repeatability, and has been widely applied in the range finding. Non-contact distance measurement can be achieved either by Time of flight (TOF) or laser triangulation [3]. Our choice is the laser triangulation system considering the complexity of the implementation and the cost. For the highway pavement inspection, the displacement sensor is desirable to be able to do: 1) Be portable (mounted on vehicle) 2) Detect moving object (up to 70 mile/h) 3) Measure the displacement of +/- 3 inches with standoff distance 12 inch 4) Work outside exposed to sunlight 5) Achieve 20 mils resolution 2. SYSTEM OVERVIEW AND ARTITECTURE The laser sensor consists of the optical system and electronic system. The components of the sensor mainly include the laser diode, the light sensitive detector, electronic subsystem. As show in fig. 1, the electronic system of the sensor is composed of several parts including Position sensitive detector (PSD), current to voltage conversion circuit, background suppression circuit, and laser power intensity control circuit. The outputs of PSD are current signals which need to be converted to voltage signals for further signal processing. The background suppression circuits are composed of sample-and-hold circuits and operational amplifiers. The sum signal of two PSD outputs is set as the input to the PID controller for adjusting the laser power. Laser Triangulation Configurations A common optical arrangement of the laser triangulation sensor is shown in fig. 2. The laser beam is projected onto the target through the collimator lens by laser diode. The light is scattered by the target surface and a portion of the light is collected by the convex lens and projected onto the light sensitive detector. When the position of the target changes, the path of the scattered light alters and the spot on the detector moves correspondingly. The optical components including the light sensitive detector, laser diode and the lens are well arranged to maintain the Scheimpflug condition [5] and guarantee the good focusing in the entire measuring range.

2 PSD I1 I2 I/V I/V V1 V2 Subtraction Summation Sub signal Sum signal Background Suppression (S /H ) SUB/SUM Data output DAQ Clock and Timing Laser Diode LD d rivie r Power Control PID Controller α β Fig 1 System block based on the factors of speed, resolution and implementation complexity. PSD is a pure analog device and relies on a current generated in a photodiode dividing in one or two resistive layers. Once a light strikes the PSD, an electric charge proportional to the light intensity is generated at the incident position. The charge is collected by the electrode X1 and X2 as photocurrents (see fig. 3). The relation between the light position and current at electrode X1 and X2 is given by the equation (1) [6]. I I 2X I L X 2 X 1 = X 2 + I X 1 X A (1) Wavelength Selection Fig. 2 Laser triangulation principle The maximal intensity of sun radiation is in the range between 0.1um and 0.4um. To maximally avoid the influence of sun radiation and decrease the skin and eye hazard, IR laser source is chosen over UV laser source. The longer the IR wavelength is, the less dangerous the light source is to the eyes, since human eyes absorb less light of longer wavelength [3]. In our system, we choose IR laser diode with the wavelength of 785nm. Working principle of the PSD gives the design stability and reliability. The electronics needed for signal processing is simple and can be implemented at low cost. The resolution of PSD is determined by the PSD resistance length and SNR. The resolution can be further enhanced by using the stored reference points as a look up table. Moreover, PSD outputs the analog signal that can quickly respond to the light spot change. Our sensor chooses one dimensional PSD S1352 from Hamamatsu. It has photo sensitivity of 0.55A/w at the wavelength 785nm which is close to the peak sensitivity wavelength of the detector and has the active area of 2.5 x 32 mm which satisfies the desired measuring range. Light Sensitive Detector Two commonly used light sensitive detectors in laser displacement sensors are Charged Coupled Device (CCD) and Position Sensitive Device (PSD). PSD is chosen as the light sensitive detector in our displacement sensor Fig. 3 PSD section view [6]

3 Background Suppression The sunlight has the wide range of the spectrum from 100nm to 10 6 nm that covers the sensitive wavelength of PSD. To suppress the influence background like sunlight and fluorescent light, three main methods could be implemented [3]. 1) Modulated laser 2) Optical filter 3) Background subtraction All those three methods mentioned above are implemented in our sensor. Since the central wavelength of the laser diode is 785nm, a narrow band pass optical filter with the center wavelength of 785nm and the full width at half maximum transmission (FWHM) of 5 nm is used in the optical subsystem. But the narrow band pass filter has limited acceptance of the incident angle and it reduces the intensity of the light through the filter, therefore reduces SNR. This point needs to be taken into account when implementing the optical filter in laser triangulation sensor. Though the optical filter is used, the background can only be suppressed and can not be completely eliminated. Only by background subtraction can the influence of the background light be removed. Therefore, the laser is modulated for background subtraction. frequency of our system is KHz, which fully satisfy the requirement of the horizontal resolution. PID Control Circuit Though PSD is the dominant technology in laser displacement sensor because of its advantages, for example, high resolution and fast response, its inherent characteristic that it finds the position of the center of the average light quality distribution rather than the peak value of the light distribution results in that PSD output varies depending on the incident light power. The power received by the PSD significantly depends on the distance to the target, the target surface characteristics, and the angle of the incident laser beam. To minimize the error caused by the light intensity, PID feedback control is implemented in our system. The sum signal of two PSD outputs reflects the power of incident light and is fed to PID control circuit. After compared with the reference value, the difference is sent to the PID circuit which adjusts the power of laser diode. Before the PID circuit is implemented in our system, some simulation work is done to investigate into the dynamical range of the PID controller. 3. SIMULATION AND EXPERIMENT RESULT PID Control Simulation in Simulink According to the inverse-square law, the power of the light PSD receives is inversely proportional to the square of the distance to the light source and is proportional to the light source power. For investigation into the dynamic range and the error of the PID controller, experiment data that describe the relation between the power received by PSD, distance and light source power were collected and a model based on polynomial fitting is done. Fig. 5 shows the experiment data, the modeled data and the error distribution. Fig. 4 Background subtraction circuit The method of eliminating the background is shown in fig. 4. The output of PSD is sent to two sample-andhold (S/H) circuits simultaneously. When laser is off, PSD output which reflects the background is sampled and held. Then laser is on, PSD output which includes both laser light and background information is sampled and held. Through the amplifier, background is subtracted and eliminated. As for the modulation frequency for the laser diode, we have to consider the horizontal resolution requirement of the pavement texture inspection. The minimal interval between two sample points is 1mm. The highest speed of the vehicle which has the laser device mounted on it is 70miles/h, thus the minimal modulation frequency required for the laser diode is 32 KHz. The modulation (a)

4 signal to PID controller is able to follow the reference value. The maximal power for driving the laser diode is also known through the simulation. However, the data for model is collected under the condition that the target surface is bright and smooth. Therefore, if the surface characteristics change, the power required for reaching the set value varies. Experimental Result of PID Controller (b) After the simulation in Simulink, PID control circuit is implemented using amplifiers. INA105 is the differential amplifier that has very high precision. The difference of feedback signal and reference value is sent to the PID controller composed of operational amplifier AD825 and passive components. The output of AD825 is sent to the laser diode driving circuit. Fig. 7 PID controller (c) Fig. 5 Experimental data and model data (a) Experimental data. (b) Modeled data (c) Error distribution The error distribution shows that the model is suitable for the simulation of PID controller. And the model is implemented in PID controller in Simulink. The reference value for PID controller is 7v. The step function covers the whole measuring range of our sensor. As shown in fig. 6, the signal power_out that is the control The PID circuit is implemented in the system, and the test is done in the lab. The target is placed on the motorized bench, 9 inches away from the instrument. The reference value is set to 2.81V. Then the target moves to 15 inches away from the device. The feedback signal at the negative terminal of INA105 is acquired through the data acquisition card and the result is shown in fig. 8. Fig. 8 Feedback signal with distance change Fig. 6 Simulation of PID controller For the whole measuring range, the feedback signal follows well the set value if we ignored the noise coupled on the sampled signal. That is to say, the PID controller in our instrument helps overcome the distance issue. However, the feedback signal might not follow the reference value even with PID controller if the target surface has very low reflectivity and the power required

5 for eliminating the error of PID controller is beyond the capability of the laser diode. Sensor Calibration and Test Result The sensor generates the voltage signal. To establish the relationship between the distance and the output signal, the sensor calibration is done in the lab. The target surface has the similar texture of the pavement we are interested in. The target is moved away from the sensor in the measuring range, and the sensor output is acquired every 0.01inch. Based on the collected data, the polynomial fitting function of order 3 or 4 is obtained. In order to evaluate the stability of the system, for fixed distance, the sensor output is sampled 100 times, and the standard deviation of those data is calculated. The standard deviation also determines the resolution of our system. The standard deviation curve with respect to the distance is plotted in fig. 9. overcomes the influence of ambient light and has been successfully used in the field test. Further improvement will be to further minimize the output error caused by the light intensity and further enhance the SNR. Fig. 10 MPD comparison (courtesy TXDOT ) ACKNOWLEDGE This project is sponsored by Department of Transportation in Texas. The authors would like to acknowledge its cooperation. Thanks also to TXDOT for providing me the field test result.. Fig. 9 Standard deviation of the system (bright and smooth target) Fig. 9 shows the standard deviation in the whole measuring range for the target surface. From 9 to 12.5 inches, the standard deviation is kept below 8 mils, and then the standard deviation increases with distance. Our sensor has been successfully used in Department of Transportation in Texas for highway pavement inspection. The test result of MPD obtained by our laser displacement sensor is compared with those reference value acquired through traditional methods, sand patch method and circular track meter. Fig. 10 shows that the MPDs calculated based on the data collected by our different displacement sensors are very close to the results obtained by the traditional reference methods. The error is within the tolerance and the sensors satisfy the requirement of highway pavement texture inspection. 4. CONCLUSIONS REFERENCE [1] Bradley O. Hibbs and Roger M. Larson, Tire Pavement Noise and Safety Performance, PCC Surface Texture Technical Working Group, May 1996 [2] Henry, J.J., Evaluation of Pavement Friction Characteristics. NCHRP Synthesis 291, Transportation Research Board. Washington, D.C [3] C. Mertz, J. Kozar, J.R. Miller, and C. Thorpe, Eyesafe Laser Line Striper for Outside Use, Intelligent Vehicle Symposium, IEEE Eye-safe laser line striper for outside use, Vol 2, June 2002 Page(s): [4] Kevin K. McGhee, Gerardo W. Flintsch, Final report high-speed texture measurement of pavement, VTRC 03-R9, 2003 [5] Fumio Murakami, Accuracy Assessment of a Laser Triangulation Sensor, Instrumentation and Measurement Technology Conference, IMTC/94. [6] Hamamatsu, parts_s/s3979_etc.pdf The sensor using laser triangulation technique is discussed in this paper. The sensor achieves the requirement of the highway pavement texture inspection in terms of measuring range, speed, resolution. The sensor