A new localization method for mobile robot by data fusion of vision sensor data and motion sensor data

Pose Estimation of a Mobile Robot Based on Fusion of IMU Data and Vision Data Using an Extended Kalman Filter

“…The process of vision localization is categorised into four major steps: acquire images via camera, detect object in the current images, match the object recognised with those contained in the database and finally, calculate the pose as a function of the recognised object. In this work, a forward-looking single camera (monocular) was used because it provides a high number of markers, thus allowing good motion estimation accuracy, if the objects are closers to the camera [ 19 , 64 ].…”

“…The vision method provides orientation of the object {o} wrt to camera coordinate frame {c}, R co using the method in [ 67 ], to calculate the pose of the mobile robot with respect to the camera based on the pinhole camera model. The monocular vision positioning system used in [ 19 ], was used to estimate the 3D camera from the 2D image plane. The relationship between a point in the world frame and its projection in the image plane can be expressed as: where is a scale factor, and are homogenous coordinates on image plane and world coordinate is a projection matrix.…”

“…More discussion on markers will be presented in Section 2 . The fusion methods, such as the Kalman filter or extended Kalman filter, usually adopt iterative algorithms to deal with linear and non-linear models, and hence convergence is not always assured [ 19 , 20 ]. For an autonomous mobile robot to localize and determine its precise orientation and position, some techniques are required to tackle this problem.…”

Mobile Robot Transportation for Multiple Labware with Hybrid Pose Correction in Life Science Laboratories

“…In addition, the required time is not satisfying. Methods using multiple sensors may be efficient and stable [30], [31]. But, the sensors could affect each other if they are employed in large areas.…”