Robotics and Computer Vision Laboratory Login  
  Robotics and Computer Vision Laboratory kaist logo
Archive Courses

Publications
Home  >  Research  >  Publications

 
[Domestic Conference] Stabilizing Visual SLAM using Limited Range Measurements
Workshop on Image Processing and Image Understanding (IPIU) , February 2012
Download
  IPIU2012_StaVSLAM.pdf IPIU2012_StaVSLAM.pdf (558.8K) [138]
Abstract
One of the most important problems in incremental visual odometry is to maintain the consistent scale throughout the whole sequence. Absolute and independent scale measurements used in sensor fusion systems make the scale of the local trajectory stable and consistent. However, the effect of the absolute range measurements to the stability of the estimated trajectory has not been investigated yet. In this paper, we analyze the effect of the absolute range measurements to the trajectory estimated in the visual SLAM and sensor fusion frameworks. The analysis aims to reveal the role of the absolute measurements in the previous camera-laser sensor fusion method and how many measurements are required to stabilize the visual SLAM. We have found that the stability of the trajectory throughout the whole sequence depends on the number of measurements in the previous methods. To stabilize the visual SLAM and to improve the robustness of the sensor fusion methods we propose a scale optimization algorithm stabilizing the estimated sensor trajectory by utilizing only a few range measurements efficiently. The proposed method has been tested with indoor and outdoor sequences, and the results show that the proposed incremental method can successfully recover the sensor trajectory accurately only with a few range measurements.

 
   
 

Robotics and Computer Vision Laboratory
KAIST | Electrical Engineering | Contact Us | Sitemap