Mapping in dynamic environments using stereo vision

Looking at Vehicles on the Road: A Survey of Vision-Based Vehicle Detection, Tracking, and Behavior Analysis

“…Using curve-fitting techniques, such as the Hough transform [115] or the RANdom SAmple Consensus (RANSAC) [116], disparity can be modeled as a function of the v coordinate of the disparity map, and pixel locations can be classified as belonging to the ground surface if they fit this model [113]. The v-disparity has been widely used in stereo vision for intelligent vehicles [83], [113], [117]- [122].…”

“…In [133], the concept of 6D vision, i.e., the tracking of interest points in 3-D using Kalman filtering, along with ego-motion compensation, is used to identify moving and static objects in the scene. Optical flow is also used as a fundamental component of stereo-vision analysis of the on-road scene in [84], [117], [126], [127], [131], and [133]- [137]. A 3-D version of optical flow, in which a least squares solution to 3-D points' motion is solved, is used in [138].…”

“…In [117], the ground plane model is fit using total least squares. In [126] and [134], the ground plane is estimated using RANSAC to fit the plane parameters [116].…”

Bayesian environment representation, prediction, and criticality assessment for driver assistance systems

“…• Methods that filter out measurements supposed to belong to dynamic objects and subsequently use a traditional mapping algorithm suitable for a static world [42,100,141,183,184,193,229,[259][260][261]263].…”

“…With this laser scanner-based system, trails of moving objects can be suppressed and the tracking itself benefits from the grid mapping. In [141], the disparity image is classified into regions of stationary and moving parts by generating object hypotheses by segmenting object contours on a v-disparity image, which are tracked over time with additional exploitation of optical flow within the measurement equations. A classifier based on a sequential probability ratio test on the residuals of a stationary and a moving process model is used to distinguish moving from stationary parts of the environment.…”