Dense RGB-D visual odometry using inverse depth

Abstract

In this paper we present a dense visual odometry system for RGB-D cameras performing both photometric and geometric error minimisation to estimate the camera motion between frames. Contrary to most works in the literature, we parametrise the geometric error by the inverse depth instead of the depth, which translates into a better fit of the distribution of the geometric error to the used robust cost functions. To improve the accuracy we propose to use a keyframe switching strategy based on a visibility criteria between frames. For the comparison of our approach with state-of-the-art approaches we use the popular datasets from the TUM for RGB-D benchmarking as well as two synthetic datasets. Our approach shows to be competitive with state-of-theart methods in terms of drift in meters per second, even compared to methods performing loop closure too. When comparing to approaches performing pure odometry like ours, our method outperforms them in the majority of the tested datasets. Additionally we show that our approach is able to work in real time and we provide a qualitative evaluation on our own sequences showing a low drift in the 3D reconstructions. We have implemented this method within the scope of PCL (Point Cloud Library) as a branch of the code for large scale KinectFusion, where the original ICP system for odometry estimation has been completely substituted by our method. A PCL fork including the modified method is available for download .

RGB-D Odometry and SLAM

“…-Camera tracking: The camera tracking thread estimates the incremental camera motion. The most simple approach is to use the frame-to-frame constraints (e.g., [36,55]). This is in fact inevitable when bootstrapping the system from the first two views.…”

Probabilistic RGB-D odometry based on points, lines and planes under depth uncertainty

“…The dense RGB-D odometry method, termed DVO [16], which is based on the minimization of the photometric and geometric error, in [17], has also been improved, in [18,19], by considering the depth error. [18] proposed using the inverse depth to parameterize the geometric error, whereas [19] proposed using the image derivatives to weight the residuals of the minimization problem.…”

Low Cost Multi-Sensor Robot Laser Scanning System and Its Accuracy Investigations for Indoor Mapping Application

“…The proposed robot system is designed to include 2D/3D imaging sensors (Laser scanner/ High resolution panoramic camera), position/orientation sensors (Odometer/MEMS-IMU (Micro-Electro-Mechanical SystemInertial Measurement Unit)) and two wheel differential chassis operating on ROS (Robot Operating System, www.ros.org/) to realize indoor mapping data acquisition without human intervention. The robot mobile mapping system can be applied to applications such as indoor scene visualization (Camplani et al, 2013), floor plan generation (Choi et al, 2015), BIM (Building Information Model) construction (Xiong et al, 2015), simulation (Gemignani et al, 2016), indoor navigation (Gutierrez-Gomez et al, 2016), virtual reality (Seibert et al, 2017) and etc.…”