Items with full text/Total items : 90429/105609 (86%)
Visitors : 10283261
Online Users : 103
Please use this identifier to cite or link to this item:
|Title: ||Wireless stereo vision for an omnidirectional robot|
|Authors: ||張漢威;Zhang, Han-Wei;蔡志仁;Tsai, Zhi-Ren;*;林進燈;Lin, Chin-Teng|
|Issue Date: ||2018-10-22|
|Abstract: ||Introduction: Image processing has been implemented for numerous applications but is yet to be robustly and
practically applied for the control of omnidirectional robots. This application uniquely requires a visual servo
function and a model-free design; stereo vision, which involves extracting three-dimensional information from
images from multiple viewpoints, can be used to meet these requirements. Methods: Here, we propose a
prototype of a two-camera system to enable wireless stereo vision. This system uses a controller to locate
light-emitting diode markers placed on an omnidirectional robot as a robust target to guide the robot through
various obstacles. This wireless system includes a control system that leverages artificial intelligence to
recognize itself via remote stereo vision and adapt to its environment accordingly. However, this scenario
introduces many control problems including time-varying delay, nonlinearity, and uncertainty. Hence, given
the sampling period of this control system, a model-free method is proposed. A visual-delay servo is used to
guide the robot and an adaptive prediction-control law is implemented via an island-type neural network.
Moreover, the sampling time is accounted for in the continuous-time system to design the digital controller.
Results: The proposed system is tested for guiding a robot into a garage. The experimental results show that
the proposed autonomous navigation, stereo vision servo, and CIE-Lab color transformation enable a rapid
and accurate pose acquisition for this robot. Moreover, the proposed technique provides better performance
compared to traditional methods. Conclusion: These results demonstrate the potential of this system, which
does not rely on physical models, to be used for a wide range of applications in the future.
Keywords: Wireless stereo vision; Visual-delay servo; LED-based marker; CIE-Lab; Island-type neural
|Relation: ||JOURNAL OF ENGINEERING TECHNOLOGY|
|Appears in Collections:||[資訊工程學系] 期刊論文|
Files in This Item:
All items in ASIAIR are protected by copyright, with all rights reserved.