Acta Mechanica Slovaca, 3/2008 1 TESTING AND CALIBRATION OF IR PROXIMITY SENSORS Václav KRYS, Tomáš KOT, Ján BABJAK, Vladimír MOSTÝN Testování a kalibrace IR snímačů vzdálenosti Na katedře robototechniky fakulty strojní vysoké školy báňské se věnujeme návrhu a realizaci několika servisních robotů. Jedním z hlavních požadavků na tyto robotické systémy je schopnost orientace v neznámém pracovním prostředí. Aby bylo možno tyto požadavky splnit, je potřeba robota vybavit množstvím senzorů. Jedním z nejčastěji používaných principů, je senzor na bázi detekce odrazu infračerveného záření od překážky. Pokud ale chceme nasadit tyto senzory v praxi je potřeba ověřit jejich funkci, stanovit parametry překážek tak aby byla zajištěna požadovaná přesnost a být schopen kalibrovat jejich výslednou hodnotu tak aby naměřený údaj co nejvíce odpovídal skutečnosti. Článek popisuje proces měření a kalibrace senzorů SHARP GP2D120 a GP2Y0A02, zakoupených naší katedrou. Testování probíhalo v reálných podmínkách za použití mobilního robotu, kterým katedra disponuje. Klíčová slova: IR, senzor, vzdálenost, kalibrace, mobilní robot On the Department of robotics at VŠB we engage in design and realization of several service robots. One of the main functions is the ability to navigate in unknown environment. To meet these requirements, it is necessary to equip the robot with sensors. One of the most common principles of obstacle distance measuring is detection of IR radiation reflection. If we want to use such sensors in practice, we must verify their function, determine the influence of obstacle material and calibrate the sensor so that the measured value will correspond to the actual distance. The article describes the process of testing and calibration of SHARP sensors, purchased by our department. The testing was performed in real conditions using a mobile robot. Keywords: IR, sensor, distance, calibration, mobile robot Ing. KRYS, Václav, Ing. KOT, Tomáš, Ing. BABJAK, Ján, prof. Dr. Ing. MOSTÝN, Vladimír, VŠB Technická Univerzita Ostrava, Fakulta strojní, Katedra robototechniky, Ostrava Recenzent:
2 Acta Mechanica Slovaca, 3/2008 1 INTRODUCTION The ability to detect obstacles is one of the most important requirements an autonomous mobile robot should meet. That s why mobile robots usually are equipped with a variety of sensors. One of the basic object detectors used in robotics are IR sensors. There are many types of them, varying in their parameters and price. Important parameters are measuring distance range and precision. Also on the Department of robotics was developed a sensor working on the basis of detection of IR light reflected on an obstacle [1]. The Sharp Company engages in production of much more sophisticated IR sensors and offers a full range of sensors for a huge variety of applications. Our department bought two types of these sensors (GP2D120 and GP2Y0A02, see Fig. 1) for testing purposes to prove their capability of being used on mobile robots developed by the department. Fig. 1 Sharp IR sensors GP2D120 and GP2Y0A02 2 TESTING DEVICE The measuring distance range is given by two values the minimal and the maximal range. It is not possible to use just one long range sensor, because it would be not able to scan the area near the robot. Our testing device consists of two pieces of GP2D120 sensors with the range of 4 to 30 cm and one GP2Y0A02 sensor with the range of 20 to 150 cm. The sensors are mounted in a row on a plastic bracket, the one with the higher range in the middle and the other two on the sides. The bracket is intended to be placed on the front side of a mobile robot (see Fig. 2). Because of application of three sensors, it is possible to detect obstacles in the whole range of distances and also to detect how the obstacle is positioned in relation to the robot.
Acta Mechanica Slovaca, 3/2008 3 Fig. 2 Testing bracket with IR sensors mounted on a mobile robot The sensors return the measured distance in a form of analog signal. For its further processing it is necessary to use an AD converter and an interface, which would send the values to the PC for utilization in the robot s control system. All these functions are arranged by a microprocessor Atmel ATMega8, which contains both the AD converter and RS232 interface for communication with a PC. Testing Device (mobile robot) Wireless data transfer High level control system GP2D120 GP2Y0A02 GP2D120 ATMEL ATMega8 Radiomodem 433MHz Radiomodem 433MHz Fig. 3 Scheme of the distance measuring subsystem
4 Acta Mechanica Slovaca, 3/2008 3 MEASURING PROCEDURE The dependence between voltage output and corresponding distance for the used sensors is not linear (Fig. 4). It is necessary to linearize it to be able to read distance values in the control system. Fig. 4 Measuring characteristics of the GP2D120 sensor MS Excel can be used to find out the equation of trend line for the graph. In the case of GP2D120 sensor, the power interpolation function is: 0,9283 U 10,49 D, (1) where U is the output voltage in volts and D is the corresponding distance in centimeters. To compute distance based on voltage, it is necessary to make inverse function, which has the following form: 1,077 10,49 D (2) U This function can be easily applied in the control system, but it is not very accurate in the whole range, because the trend line does not fit perfectly. Thus it was desirable to make a different method of calculation. We found out that a good method is to simply make a table of points describing the active portion of the graph (Fig. 4) and perform a simple linear interpolation between those points.
Acta Mechanica Slovaca, 3/2008 5 Voltage [V] Distance [cm] 3,00 3,5 2,80 4,0 1,60 8,0 1,10 12,0 0,85 16,0 0,70 20,0 0,60 24,0 0,50 28,0 0,40 32,0 0,35 36,0 0,30 40,0 Tab. 1 Control points for measuring characteristics of the GP2D120 sensor Provided we have three sensors in the described configuration, we are able to detect also obstacles that extend only to a part of the robot s front side and more importantly the control system has the information about presence of a short obstacle and can use it for navigation purposes. Also it is able to compute the angle of approach to a wall. Obstacle Obstacle Obstacle Mounting bracket Mounting bracket mounting bracket Fig. 4 Basic types of obstacles detectable by the system 4 CALIBRATION During the testing, we noticed that the sensors outputs are not in perfect accordance with the measuring characteristics graphs. Even two pieces of the same type of sensor (in our case the two sensors with lower range) return different values for the same actual obstacle distance. Thus it was necessary to perform some calibration. A simple mechanism was built, allowing to position a reference obstacle to specified distances from the sensor. This way it was possible to make corrections to the control points table and to make separate tables for each individual sensor.
6 Acta Mechanica Slovaca, 3/2008 5 CONCLUSION Outcome of this project is a universal obstacle detection device for mobile robots, which can be used on any robots developed by our department. The device is capable of measuring distance of obstacles in the overall range of 4 to 150 centimeters and to detect obstacles extending only to a part of the robot s front side. The control system can also calculate robot s angle of approach to a wall. The device increases the abilities of a mobile robot to navigate in unknown environment and will be used not only on operator controlled robots but in the future also on autonomous robots with a neural net based control system. Acknowledgements This article was compiled as part of projects FT-TA3/014, supported by the Fund for University Development from the Ministry of Industry and Trade. REFERENCES [1] NOVÁK, Petr, BABJAK, Ján, TVARŮŢKA, Adam. Intelligent unit of the IR proximity sensors with I2C interface. Acta Mechanica Slovaca. 2006, vol. 10. ISSN 1335-2392..