Study of Navigation Algorithms Based on Stereo Vision for Autonomous Cart
碩士 === 國立彰化師範大學 === 電機工程學系 === 99 === Abstract This thesis proposes matching algorithms for binocular images to generate the disparity map used for autonomous cart obstacle avoidance navigation. The navigation control algorithms including corresponding binocular image point matching, flat road detec...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Published: |
2011
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/13203019432834043183 |
| Summary: | 碩士 === 國立彰化師範大學 === 電機工程學系 === 99 === Abstract
This thesis proposes matching algorithms for binocular images to generate the disparity map used for autonomous cart obstacle avoidance navigation. The navigation control algorithms including corresponding binocular image point matching, flat road detection and path planning. The optical axis of two identical cameras mounted on an autonomous cart is arranged parallel to one another. After the epipolar lines of the binocular images are calibrated at the same horizontal line, the corresponding points are searched along the epipolar line. To reduce the computations the image gradient is used to extract the features to find the corresponding points in the binocular images. A smoothing algorithm is used to optimize the disparity map. The flat surface for cart movement is identified and extracted using the road features on the disparity map. Using the same procedure, the width of the cart’s body in the disparity map is detected. Real-time path planning is executed to move the cart within the available space, safely avoiding obstructions and potholes. The experimental results show that the autonomous cart control algorithms have the ability to avoid static and dynamic obstacles to achieve visual navigation in complex indoor and outdoor spaces.
|
|---|