Localization is a key capability for autonomous vehicles especially in urban scenarios. We propose the use of pole-like landmarks as primary features in these environments, as they are distinct, long-term stable and can be detected reliably with a stereo camera system. Furthermore, the resulting map representation is memory efficient, allowing for easy storage and on-line updates. The localization is performed in real-time by a stereo camera system as a main sensor, using vehicle odometry and an off-the-shelf GPS as secondary information sources. Localization is performed by a particle filter approach, coupled with an Kalman filter for robustness and sensor fusion. This leads to a lateral accuracy below 20 cm in various urban test areas. The system has been included in our autonomous test vehicle and successfully demonstrated the full loop from mapping to autonomous driving.

Semi-Global Matching (SGM) is widely used for real-time stereo vision in the automotive context. Despite its popularity, only implementations using reconfigurable hardware (FPGA) or graphics hardware (GPU) achieve high enough frame rates for intelligent vehicles. Existing real-time implementations for general purpose PCs use image and disparity sub-sampling at the expense of matching quality. We study methods to improve the efficiency of SGM on general purpose PCs, through fine grained parallelization and usage of multiple cores. The different approaches are evaluated on the KITTI benchmark, which provides real imagery with LIDAR ground truth. The system is able to compute disparity maps of VGA image pairs with a disparity range of 128 values at more than 16 Hz. The approach is scalable to the number of available cores and portable to embedded processors.

This paper presents an approach to the problem of on-line stereo self-calibration. After a short introduction of the general method, we propose a new one, based on the minimization of matching costs. We furthermore show that the number of matched pixels can be used as a quality measure. A Metropolis algorithm based Monte-Carlo scheme is employed to reliably minimize the costs. We present experimental results in the context of automotive stereo with different matching algorithms. These show the effectiveness for the calibration of roll and pitch angle offsets.

Automotive applications based on stereo vision require robust and fast matching algorithms, which makes semi-global matching (SGM) a popular method in this field. Typically the Census transform is used as a cost function, since it is advantageous for outdoor scenes. We propose an extension based on center-symmetric local binary patterns, which allows better efficiency and higher matching quality. Our second contribution exploits knowledge about the three-dimensional structure of the scene to selectively enforce the smoothness constraints of SGM. It is shown that information about surface normals can be easily integrated by weighing the paths according to the gradient of the disparity. The different approaches are evaluated on the KITTI benchmark, which provides real imagery with LIDAR ground truth. The results indicate improved performance compared to state-of-the-art SGM based algorithms.

Trajectories extracted from an object tracker are a valuable source of information to derive describing parameters for traffic situation analysis. In order to ensure the quality of these parameters, the tracker's output has to be evaluated. This is a challenging task, since the choice of the metrics and the choice of the matching --- of ground truth tracks and the output of the tracker --- influences the results considerably. We present the evaluation of a Kalman-filter based tracker using ground truth data of traffic scenes. The Kalman-Filter is object space based and uses observations obtained from multiple camera views. In the evaluation, a two level approach is employed. The first level assesses the ability to correctly locate the objects. The second one evaluates the consistent identification of objects throughout the scene. Matching of tracks is done using a spatial overlap measure. Subsequently, several metrics are applied, whose properties are outlined. The results for simulated and real traffic data are presented. Finally, the implications of aggregation and normalization of these metrics in order to estimate the algorithm's performance are discussed.