Laser-Inertial Aided 30 Scanner Using Geometric Invariant for Terrain Construction

M. S. Hendriyawan, Achmad and Mohd Razali, Daud and Pebrianti, Dwi and Saifudin, Razali (2015) Laser-Inertial Aided 30 Scanner Using Geometric Invariant for Terrain Construction. In: 2nd International Conference on Mechanics and Mechatronics Research (ICMMR2015) , 11-12 June 2015 , Kuala Lumpur, Malaysia. pp. 957-963., 799.


Download (5MB) | Preview


Researchers in robotic vision technology are facing larger challenges, where the 2D technology has flaws in complex robot navigation in 3D space. Using 3D scanner, the robot is able to get a more detailed terrain construction, making it easier to carry out its tasks. The 3D image is obtained by fusing the Hokuyo URG-04LX and the 6-DOF IMU that consists of acceleration sensor and gyro sensor. IMU sensor outputs are the angle, speed, and position in 3D. Nevertheless, just the value of the angle is used in this study to construct 3D images based on geometric invariant. To reduce the interference in the sensor output, two types of filter are applied; the Gaussian filter used on the output of 2D LRF, while the complementary filter is applied to the output of the IMU sensor. Angle measurement plays an important role in term of geometric invariant for terrain construction. The complementary filter has provided the best angle measurement results with the lowest error on time constant (-r) = 0.475s and sampling time (dt) = lOms. Thus, the proposed systems have successfully made an obvious 3D image of the terrain in the indoor testing.

Item Type: Conference or Workshop Item (Speech)
Uncontrolled Keywords: 30 scanner, geometric invariant, Hokuyo URG-04LX, IMU sensor, complementary filter, gaussian filters
Subjects: T Technology > TK Electrical engineering. Electronics Nuclear engineering
Faculty/Division: Faculty of Electrical & Electronic Engineering
Depositing User: Ms. Nurul Hamira Abd Razak
Date Deposited: 12 Jan 2017 02:16
Last Modified: 16 Nov 2017 01:28
Download Statistic: View Download Statistics

Actions (login required)

View Item View Item