PERANCANGAN MODUL LOCALIZATION UNTUK MOBILE ROBOT BERBASIS SENSOR INERTIAL MEASUREMENT UNIT DENGAN KALMAN FILTER

DIAR FAHRUDDIN SASONGKO, Adha Imam Cahyadi, S.T., M. Eng, D.Eng.; Ir., Samiadji Herdjunanto, M.Sc.

2014 | Skripsi | TEKNIK ELEKTRO

Saat ini, teknologi Mobile Robot telah berkembang pesat. Terlebih lagi, cakupan dan aplikasinya semakin luas di kehidupan sehari-hari. Kemampuan utama pada Mobile Robot sangat bergantung pada navigasi ketika berpindah posisi pada suatu lingkungan. Agar dapat melakukan navigasi yang benar, Mobile Robot perlu suatu localization terhadap lingkungannya. Salah satu metode penentuan posisi secara lokal adalah Inertial Navigation System (INS) yang menggunakan sensor Gyroscope dan Accelerometer yang dipakai dalam Skripsi ini. Pada metode ini, data dari sensor Gyroscope dan Accelerometer berada dalam frame sensor, sehingga diperlukan DCM untuk mentransformasikannya agar menjadi dalam frame bumi. Kalman Filter digunakan untuk mengurangi derau yang ada pada raw data dari sensor Gyroscope dan Accelerometer. Kompensator Gravitasi digunakan untuk menghilangkan pengaruh percepatan gravitasi yang dibaca sensor Accelerometer. Sehingga, percepatan linear dari data Accelerometer merupakan benar-benar percepatan linear sensor saat bergerak. Saat sensor dalam keadaan diam, hasil perhitungan mengalami pergeseran nilai (drift) yang cukup signifikan karena percepatan linear dan kecepatan angular sensor tidak tepat bernilai 0. Untuk mengatasinya, digunakan Zero Velocity Compensator saat sensor benar-benar dalam keadaan diam. Hasil pengujian yang didapatkan menunjukkan bahwa algoritma localization yang digunakan sudah cukup akurat. Ketika sensor digerakkan dari posisi (X; Y) = (0; 0) cm ke (X; Y) = (10; 10) cm dengan Yaw = 0, hasil localization yang didapatkan adalah posisi (X; Y) = (6,97; 9,95) cm, Yaw = -0,1643 rad, IAE (X; Y) = (4973; 4967), dan IAE Yaw = 850. Ketidakakuratan tersebut disebabkan oleh beberapa hal, di antaranya kesulitan menentukan nilai matriks Process Noise Covariance (R) dan Measurement Noise Covariance (Q) yang tepat pada perhitungan Kalman Filter. Terlebih lagi, karakteristik sensor juga tidak ideal.

Currently, technology for mobile robot has been developed rapidly. Moreover, its scope and application have also been developed widely in a daily life. The ability of mobile robot is highly dependent on the navigation while moving in arbitray environment. In order to perform correct navigation, mobile robot needs a localization about its environment. One of the methods to determine the local position is Inertial Navigation System (INS) that uses the Gyroscope and Accelerometer sensor that is adopted in this thesis. In this work, data from the Gyroscope and Accelerometer sensors are in the sensor frame, so it is necessary to use the DCM to transformate it to be in the earth frame. Kalman filter is used to reduce the existing noise in the raw data from the Gyroscope and Accelerometer sensor. Gravity compensator is also used to eliminate the influence of gravity acceleration that Accelerometer sensor reads. Thus, the linear acceleration from Accelerometer sensor data is truly a sensor linear acceleration while it is moving. When the sensor is not moving, the result of the calculation has a quite significant drift on the sensor position and direction due to sensor linear acceleration and angular velocity are not exactly 0. To correct this, we use Zero Velocity Compensator when the sensor is actually not moving. The results show that the algorithm of localization is quite accurate. When the sensors are moved from (X; Y) = (0; 0) cm to (X; Y) = (10; 10) cm with constant Yaw = 0, the results of localization are (X; Y) = (6,97; 9,95) cm, Yaw = -0,1643 rad, IAE (X; Y) = (4973; 4967), and IAE Yaw = 850. The inaccuracy is due to the difficulty to determine the value of Process Noise Covariance (R) and Measurement Noise Covariance (Q) matrices appropriately for the calculation of the Kalman Filter. Moreover, the characteristics of sensor are not ideal aswell.

Kata Kunci : Localization, DCM, Sensor IMU, Kalman Filter, Kompensator Gravitasi, Zero Velocity Compensator (ZVC)