JP Journal of Heat and Mass Transfer
Special Volume, Issue II, Advances in Mechanical System and ICT-convergence, Pages 221 - 237
(July 2018) http://dx.doi.org/10.17654/HMSI218221 |
|
AN EFFECTIVE DEPTH MAP NAVIGATION FOR MOBILE ROBOT IN INDOOR ENVIRONMENTS
Dang Khanh Hoa, Than Viet Duc, Vu Minh Hoang, Vu Song Tung and Nguyen Tien Dzung
|
Abstract: This article presents a solution of navigation in an indoor environment for a mobile robot that only uses a depth stream. In this work, a real-time navigation method is integrated in the self-built robotic system fitted with a Kinect sensor. First, the method applies a viable mobility principle that is a road vehicle only be able to travel on the ground. So, this method starts with finding the ground from the depth video which is provided steady by Kinect. Based on shapes of this reliable ground plane, a robust moving direction selection algorithm is proposed. In the meantime, the robot system always follows the minimum path distance strategy to the given target. At the same time, with a predetermined map of operating environment, robot navigation problem was resolved with the same successful rate of recent methods such as Fast Sampling Plane Filtering (FSPF) method. The robotic system is navigated more robust and refined than the neural network method with a 10-step adjustable angle. Experiments have shown that the stability and accuracy of the proposed navigation algorithm is up to 98%. The processing time of the navigation proposed method is minimized to meet the robot’s real-time applications. |
Keywords and phrases: depth map, ground plane, path direction, navigation. |
|
Number of Downloads: 374 | Number of Views: 3428 |
|