occupancy grid mapping wiht ray potentails

This article proposes a novel approach for grid object representation that can be used with a sparse scene description.

The data collection and labelling will contribute to the overall cost of algorithm verification.

The aforementioned methods of occupancy grid evaluation require accurate and dense ground truth. This limits their application to only well-described test scenarios or to precisely labelled experimental data. In the object extraction steps of these methods, the occupancy grid must be carefully analysed and compared with accurate scene descriptions. In the literature, there are metrics for object tracking and for collision risk estimation. This category of quality measurement is suited for specific grid applications. The third type of occupancy grid evaluation technique is the detection and monitoring of specific grid object representations. This approach is useful in grid development nonetheless, it already contains a decision-making binarisation process, which may alter the results of the assessment. Another evaluation method for the occupancy grids is using cell-wise binary comparisons with the reference maps. The evaluation of a map is currently often carried out by visual inspection, leading to highly subjective results. Based on the output grid, components such as localisation, path-planning and scene recognition make decisions about future vehicle actions.īoth the development of the occupancy grid algorithm and the decision-making process require a measure of quality for the computed grid maps. In the automotive perception software stack, the occupancy grid is an intermediate step, which combines and accumulates sensor information over time and applies different methods to filter noise. The task of this environmental model is to calculate the probability that each grid cell is either occupied or free based on sensor observations. The occupancy grid is a multi-dimensional spatial lattice, where each cell stands for an independent portion of space. One of the most popular modelling frameworks is the occupancy grid map.

Almost all automotive perception systems require a stationary representation of the environment around the vehicle.

Occupancy grid mapping wiht ray potentails