SEMANTIC SEGMENTATION MODEL FOR MOBILE ROBOT’S NAVIGATION WITH OPTIMAL ARCHITECTURE OF RESOURCE

Abstract

Due to the wealth of information extracted from digital images, computer
vision plays a significant role in mobile robot navigation. Based on the captured
images, mobile robots localize and move to the intended destination. Due to the
complexity of the environment, obstacle avoidance still requires a complex sensor
system with a high computational efficiency requirement. This study offers a real-
time solution to the problem of extracting corridor scenes from a single image.
Using ultra fast semantic segmentation model integrating with quatization
technique to reduce the numerous of training parameters and computational cost.
Moreover, the mean Intersection over Union (mIoU) achieves 89% and high
accuracy is 98%. The simulation results are compared with recent methods to
prove the feasibility of the prosed method. Finally, the authors effectively apply
the segmented image to construct the mobile robot frontal view to identify the
available mobile robot path planning.