Strategic navigation for robotic wheelchair based on geometrical information of augmented reality markers

Abstract

This paper proposes a mobile robot navigation method based on augmented reality technique using not only AR markers’ patterns but their configurations. Mobile robot requires self-location to navigate toward destination. The authors propose a navigation system for wheelchairs that uses position and orientation of AR markers as well as information connoted in the AR marker patterns. AR technology is applied to estimate relative location of robot. The system consists of three elements: AR markers, a web camera, and a motor-driven wheelchair, WHILL Model CR. When the camera on the wheelchair recognizes an AR marker, it creates an action command according to both ID of the recognized AR pattern and arrangement of the AR marker. Identified position and posture of the AR markers are used to calculate the complement rotation angle to adjust the wheelchair to the planned route. The authors have designed the techniques to plan a path for a motorized nonholonomic wheelchair to enter a narrow corridor where nonholonomic robots cannot easily drive only by turning or circular routing. The techniques calculate the radius of the arc trajectory to enter a narrow corridor based on the recognized AR marker position and posture information. Experiments are conducted to guide a robot with a width of 0.55[m] by AR markers to enter a corridor with a width of 0.7[m]. The results confirm the effectiveness of our proposal because the error of the final arrival point is 5.7[mm] from the ideal position.