In this paper, we introduce a novel method for the estimation of vehicle pose and extent in traffic surveillance scenarios based on camera data. The state estimation is performed in a common world frame, enabling the seamless integration of the image data from different viewpoints. Our approach incorporates the non-linear transformation between the measurements and the states directly into the framework of an Unscented Kalman filter. Two measurement models are proposed: one designed for bounding boxes and another for discretized object contours extracted from segmentation masks. The method is evaluated using data from a real-world traffic surveillance system, demonstrating the high effectiveness and good feasibility of our approach for localizing passing cars.

In this paper, we present a novel design for a multi-camera tracking system with occlusion-handling capabilities and its application to a highway traffic surveillance system. The fundamental concept follows the tracking-by-detection principle with monocular detectors and an LMB tracker for tracking the objects in the world frame. All data from the multi-view setup is combined into one consistent representation of the real-time traffic situation. In order to assess the inter-target occlusion scenarios in 3D, the vehicles are modeled as cuboids and their extents are estimated from the bounding boxes provided by the detectors. We re-transform the 3D occlusion estimation problem into the 2D camera space and present two methods for quantifying the occlusion state of the objects. Moreover, we propose a modification to the computation of the existence probability of undetected and occluded targets. Based on this, the tracking system is extended by an occlusion-aware detection model. We evaluate our occlusion-handling approach on a real-world traffic dataset from the Providentia++ project and show an improved tracking performance. We find that the number of misdetected targets is reduced and more track identities are preserved.

Conditional imitation learning (CIL) trains deep neural networks, in an end-to-end manner, to mimic human driving. This approach has demonstrated suitable vehicle control when following roads, avoiding obstacles, or taking specific turns at intersections to reach a destination. Unfortunately, performance dramatically decreases when deployed to unseen environments and is inconsistent against varying weather conditions. Most importantly, the current CIL fails to avoid static road blockages. In this work, we propose a solution to those deficiencies. First, we fuse the laser scanner with the regular camera streams, at the features level, to overcome the generalization and consistency challenges. Second, we introduce a new efficient Occupancy Grid Mapping (OGM) method along with new algorithms for road blockages avoidance and global route planning. Consequently, our proposed method dynamically detects partial and full road blockages, and guides the controlled vehicle to another route to reach the destination. Following the original CIL work, we demonstrated the effectiveness of our proposal on CARLA simulator urban driving benchmark. Our experiments showed that our model improved consistency against weather conditions by four times and autonomous driving success rate generalization by 52%. Furthermore, our global route planner improved the driving success rate by 37%. Our proposed road blockages avoidance algorithm improved the driving success rate by 27%. Finally, the average kilometers traveled before a collision with a static object increased by 1.5 times. The main source code can be reached at this web page:https://heshameraqi.github.io/dynamic_cil_autonomous_driving.

Point clouds generated by Lidar sensors provide detailed information about the geometry of the environment. Yet they lack semantic information, which is paramount for the choice of choosing the appropriate modelling, e.g. within a tracking system. This contribution tries to provide semantic information in the form of stationary and dynamic classification by applying a transferable belief model. In particular, we build an occupancy grid representation of the environment and correct it with a tailored transferable belief model that accounts for inconsistencies as well as non-local features. Based on these results we define a classifier that distinguishes between stationary and dynamic cells. The presented algorithm is evaluated qualitatively on real-world Valeo Scala LIDAR data and quantitatively based on an IPG Carmaker simulation.

In this paper a novel representation of the contour of an spatially extended object is proposed, which is tailored to vehicles with an unknown size and orientation that are tracked based on measurements from an automotive Light Detection and Ranging (LIDAR). We deploy quadratic uniform periodic B-Splines to directly represent a star-convex shape approximation of the object in Cartesian space. In contrast to previous approaches that work in polar space, we introduce a new walk parameter to model the contour function of an object such that the shapes parameters are well defined and lie within the same space as the measurements. A major advantage of the approach is that a scaling of the length and the width can be performed independently by scaling the basis points of the Splines.

In this paper, an implementation of an extended target tracking filter using measurements from high-resolution automotive Radio Detection and Ranging (RADAR) is proposed. Our algorithm uses the Cartesian point measurements from the target's contour as well as the Doppler range rate provided by the RADAR to track a target vehicle's position, orientation, and translational and rotational velocities. We also apply a Gaussian Process (GP) to model the vehicle's shape. To cope with the nonlinear measurement equation, we implement an Extended Kalman Filter (EKF) and provide the necessary derivatives for the Doppler measurement. We then evaluate the effectiveness of incorporating the Doppler rate on simulations and on 2 sets of real data.