Depth-based object tracking using a Robust Gaussian Filter

Abstract

We consider the problem of model-based 3Dtracking of objects given dense depth images as input. Two difficulties preclude the application of a standard Gaussian filter to this problem. First of all, depth sensors are characterized by fat-tailed measurement noise. To address this issue, we show how a recently published robustification method for Gaussian filters can be applied to the problem at hand. Thereby, we avoid using heuristic outlier detection methods that simply reject measurements if they do not match the model. Secondly, the computational cost of the standard Gaussian filter is prohibitive due to the high-dimensional measurement, i.e. the depth image. To address this problem, we propose an approximation to reduce the computational complexity of the filter. In quantitative experiments on real data we show how our method clearly outperforms the standard Gaussian filter. Furthermore, we compare its performance to a particle-filter-based tracking method, and observe comparable computational efficiency and improved accuracy and smoothness of the estimates.

A modular framework for model-based visual tracking using edge, texture and depth features

“…With the availability of low-cost RGB-D sensors, depth map is a valuable visual cue for improving the tracking accuracy and robustness [13]. Using a Gaussian filter approach, [20] considers only the depth map for estimating the pose that best registers the observed and rendered depth maps. Within a particle filter framework, [21] uses RGB-D data and a proposal distribution to improve the energy-based observation model and reduce the number of particles.…”

Estimation of the Lateral Distance between Vehicle and Lanes Using Convolutional Neural Network and Vehicle Dynamics

“…The experiment was carried out in Chongqing Intelligent Vehicle Integrated System Test Area (i-VISTA) and part of time series were selected to analyze the method efficiency. The inclination angle of the vehicle was obtained by a gyroscope and the noise that was subjected to normal distribution was filtered out by Gaussian [47,48]. It is defined that the vehicle's inclination angle is 0 at equilibrium, and the positive was to the left and the negative was to the right.…”

“…was filtered out by Gaussian [47,48]. It is defined that the vehicle's inclination angle is 0 at equilibrium, and the positive was to the left and the negative was to the right.…”

Realtime State Estimation with Tactile and Visual Sensing. Application to Planar Manipulation

“…INTRODUCTION We are interested in providing robot manipulators with the ability to track the state of a manipulation task in realtime with ordinary hardware. Visual object tracking and detection have been widely studied [1,2,3,4] and can provide reliable estimates of object pose, especially in scenarios with limited occlusions. However, it is characteristic of robotic manipulation that the robot, the gripper, or the surrounding clutter will "get in the way" and occlude the object from the cameras.…”