FVO: floor vision aided odometry

Abstract

In many indoor scenarios, such as restaurants, laboratories, and supermarkets, the planar floors are covered with rectangular tiles. We realized that the abundant parallel lines and crossing points formed by tile joints can be used as natural features to assist indoor localization, and thus we propose a novel indoor localization method for mobile robots by fusing odometry and monocular vision. The method comprises three steps. First, the heading and location of the mobile robot are approximately estimated by odometry based on incremental encoders. Second, with the aid of a camera, the lens of which points vertically toward the floor, the odometric heading estimation can be corrected by detecting the relative angle between the robot's heading and the tile joints. Third, the odometric location estimation is corrected by detecting the perpendicular distance between the image center and the tile joints. As compared with the existing indoor localization methods, the proposed method, called floor vision aided odometry, is not only relatively low in economic cost and computational complexity, but also relatively high in accuracy and robustness. The effectiveness of this method is verified by a real-world experiment based on a differential-drive wheeled mobile robot.

A Review of Sensing Technologies for Indoor Autonomous Mobile Robots

“…The core idea of the method is how to weight multiple sensors and how to determine the correction method. Wen et al [ 156 ] combine an incremental encoder with a camera to calculate the encoder-based local position and globally update the position through the similarity of the camera to the environment. Jiang et al [ 157 ] achieved the combination of feature point cloud maps and grid maps by establishing joint error co-visual between LiDAR and the monocular camera.…”

Visual-Inertial Localization for Skid-Steering Robots with Kinematic Constraints

“…Depending on sensors used and application scenarios, different localization algorithms for skid-steering robots have been developed in recent literature. In particular, Lv, Kang, and Qin [22] proposed a method for using images for correcting headings for skid-steering robot, while requiring parallel and perpendicular lines which mainly are suitable for human-made environments. which however did not provide detailed description of how the wheel encoder's measurements were integrated.…”

Frequency-Temporal Disagreement Adaptation for Robotic Terrain Classification via Vibration in a Dynamic Environment

“…To prevent the robots from suffering from such non-geometric hazards, the mobile robots must have the ability of terrain classification [ 4 ]. For another example, the robotic pose estimation, calculated by the kinematics model, which includes the slip parameters decided by the traversing terrains, usually benefits a lot from robotic terrain classification, especially in the situation without reliable global positioning systems [ 5 , 6 , 7 , 8 ]. Apart from hazards avoidance and pose estimation, many existing works have demonstrated that the performances of many other robotic fundamental functions, such as energy savings, route planning, gait control, etc., can be improved significantly from an accurate terrain classification method [ 9 , 10 , 11 , 12 , 13 ].…”