The purpose of this project is to locate the position of a known object using the Visual Odometry (VO) technology. Some of the items used in this project are the red Radio Frequency Identification (RFID) tags .The product (a robot) is to pass within the proximity of the tags. As the robot passes, it detects the tags and with the combination of visual odometry information, measurements are made. The RFID application has two parts namely the antennae and the integrated circuit system. For the integrated circuit, a MEGA ADK kit board is used. After powering de the integrated circuit, it is used with the microcontroller. The integrated circuit is used to manipulate information and operate the radio frequency (RF) signal. The MEGA ADK kit board is used together with the id-12 innovation reader which remotely reads and wrote to the tags. This communication is achieved through the ros_lib to Arduino libraries thus enabling the location of the robotic operating system (ROS) to be identified. To get the location, the red RFID tags were identified using stereo camera Visual To Odometry technique. The project shows that it is possible for the RFID technology to be used together with Visual Odometry technique to give the cheap jerseys location of a known object.
Visual Odometry (VO) is a method used to estimate the egomotion of agents such as people, vehicles, and robots by use of one or several cameras. The process is very popular for automobiles, robots, augmented reality among other areas. Since the method resembles wheel odometry, it did not come as a surprise when Nister gave it the present name in his 2004 landmark paper. (Scarammuza D, Fraundorfer F, 2011) Wheel odometry, the principle behind estimation of vehicle’s motion is the number of wheel turns in relation to time.
Similarly, in VO, motion changes are used to estimate the vehicle’s pose, only that camera images are used in this case. Although it has a drawback of requiring well-lit conditions, VO is generally advantageous to wheel odometry. Bad conditions such as slippery surface or uneven surface are not a factor in VO (Boccadoro et al. 2010).
According to demonstrations, VO is a wholesale nfl jerseys superior method in terms of accuracy of trajectory estimates. The relative position error of this method is between 0.1 and 0.2 percent. (Scarammuza D, Fraundorfer F, 2011) Considering its high Bise accuracy, VO finds applications not only as a wheel odometry supplement but also as a core feature in Inertial Measure Units (IMU’s) and Global Positioning Systems (GBS). VO has specifically proven to be very critical in areas where it is impossible to use GBS, such as in aquatic and air environments.
The history of Visual Odometry goes back to the early 1980’s when Moravec first documented the challenge in relying on only visual input to estimate the egomotion of a car. Moravec is not only credited for the first pipeline to estimate motion, but also for the renowned Moravec Corner Detector. One of the major problems with methods used before the emergence of Moravec corner detector is to track features among frames. On the contrary, the Moravec Corner Detector method, the problem of feature drift is avoided during tracking.
A lot of research has been done on VO but most of it has relied on stereo cameras thus the name Stereo VO. Shafer and Mathies who used the binocular system to improve on Moravec’s method of detecting corners generated this method (Boccadoro et al. 2010). A number of other researchers have also subsequently improved on this approach.
The monocular VO is different from stereo VO in that the 2-D data provides the source of 3-D data and relative motion. Both Omni-directional and perspective Effect cameras have been used to estimate the positions of an object successfully.
It is obvious that technology has transformed the way we do things, from governments and institutions of learning to businesses and research through to communication and almost everything. What is even amazing is the precision with which technology can be employed to for accurate results. From this project, it is proven that the RFID technology can be successfully used to estimate the position of an object with shocking precision. This is through the combination of the methods used in Visual Odometry, stereo camera Visual Odometry, and RFID (Wareed, 2009).
Data from the Easily stereo camera was integrated cheap jerseys and used to calculate the position of the robot. As the robot moved, the stereo camera was used to track its position. The Visual Odometry method used in this case was spot on in getting image features which were represented by the red RFID tags. Even as the position of the robot was tracked in 3-D, the horizon line was also determined (Visual Odometry, 2009). The accuracy error of this system was below 1% which is a good performance for a system of its kind. Visual Odometry would find applications in mapping systems. In addition, some improvements on “place recognition” may see the method being used to tell where a robot is (Visual Odometry, 2009).
Boccadoro M etal (2010) Constrained and Quantized Kalman Filtering for an RFID Robot Localization Problem Auton Robot 29: 235-251
Scarammuza D, Fraundorfer F (2011) Visual Odometry. Part 1: The First 30 Years and Fundamentals IEEE robotics & Automation
Visual Odometry for the PR2 (2009) retrieved from http://www.willowgarage.com/blog/2009/04/23/visual-odometry-pr2?page=37
Wareed Sorour (2009) RFID System Components retrieved from http://www.rfidinregion.com/how-rfid-works/54-articles/81-rfid-system-components