Visual Odometry for Absolute Position Estimation using Template Matching on Known Environment

This study proposed a novel algorithm that can estimate the absolute position of a robot by matching environmental information provided beforehand, as well as images obtained from cameras. A robust template matching technique was adopted as an algorithm that enables image matching through optimal scale value derivation for template images without being affected by robot rotation. CoppeliaSim was adopted to apply the Kalman filter to the template matching result, and conduct an experiment to evaluate the accuracy. A simulation was conducted under the condition of a tennis court established as a simulation environment. The simulation results verified that the mean accuracy was determined to be 99.45% and 99.58% based on the length of the x-axis (base line) and the y axis (side line), respectively. Based on the simulation results, it was confirmed that the proposed method can estimate the absolute position based on known environmental information with high accuracy.A demonstration of our method can be seen in the video: https://youtu.be/EdLVYNUns2Y.