Andrei Vatavu

Stereovision-Based Multiple Object Tracking in Traffic Scenarios Using Free-Form Obstacle Delimiters and Particle Filters

In this paper we present a stereovision-based approach for tracking multiple objects in crowded environments where, typically, the road lane markings are not visible and the surrounding infrastructure is not known. The proposed technique relies on measurement data provided by an intermediate occupancy grid derived from processing a stereovision-based elevation map and on free-form object delimiters extracted from this grid. Unlike other existing methods that track rigid objects using also rigid representations, we present a particle filter-based solution for tracking visual appearance-based free-form obstacle representations. At each step, the particle state is described by two components, i.e., the objects dynamic parameters and its estimated geometry. In order to solve the high-dimensionality state-space problem, a Rao-Blackwellized particle filter is used. By accurately modeling the object geometry using the polygonal lines instead of a 3-D box and, at the same time, separating the position and speed tracking from the geometry tracking at the estimator level, the proposed solution combines the efficiency of the rigid model with the benefits of a flexible object model.

Forward collision detection using a Stereo Vision System

It is known that 93% of accidents occur because of the driver behavioral errors. Many of these can be prevented. Therefore a driver assistance system must include a collision detection component that is capable of generating warnings to prevent an imminent impact and to advise the driver in the traffic scenarios. This paper presents a forward collision warning approach based on a 3D Elevation Map provided by a Dense Stereo Vision System. We detect the obstacle delimiters and taking into account the car parameters evaluate the car trajectory and associate to it a driving tunnel. A warning is generated when the obstacle delimiters intersect the current driving tunnel at an unsafe distance. Our system is robust and works in real-time.

Omnidirectional stereo vision using fisheye lenses

This work presents an omnidirectional stereo system with a direct application to autonomous logistics. The omnidirectional system uses fisheye lenses to inspect the surroundings of the automated forklift. The lenses are mounted on the top of the vehicle to allow a 360° scene reconstruction with a single stereo pair. The system provides the list of obstacles detected around the vehicle. The reconstruction of the scene is possible via a division of the fisheye images into several rectified images. A stereo matching algorithm is applied to each pair of rectified images. The computed 3D points corresponding to the rectified pairs are unified into a single cloud. The obstacle detection module operates on the unified cloud of 3D points.

Real-time modeling of dynamic environments in traffic scenarios using a stereo-vision system

Real-time modeling of dynamic environments is one of the most demanding research problems in the field of driving assistance systems. The representation module may be affected by several factors such as occlusions, unpredictable nature of the traffic participants, wrong associations or noisy measurements. In this paper we propose two different methods for real-time modeling of dynamic environments. Both motion estimation techniques are vision-based and rely on information provided by a Digital Elevation Map. The first approach consists in determining the differences between the previous and current frames. These differences are then used for computing the speeds of the traffic participants. The second motion estimation technique consists in using a fast pairwise alignment of object delimiters that are extracted by radial scanning of the Elevation Map. The final result is a more compact polygonal map with associated static and dynamic features.

Real-time dynamic environment perception in driving scenarios using difference fronts

The environment representation is one of the main challenges of autonomous navigation. In the case of complex driving environments such as crowded city traffic scenarios, achieving satisfactory results becomes even more difficult. In this paper we propose a real-time solution for two main issues of advanced driver assistance systems: unstructured environment representation and extraction of dynamic properties of traffic participants. For the real-time environment representation we propose a solution to extract object delimiters from the traffic scenes and represent them as polygonal models. In order to track dynamic entities, an intermediate evidence map named “Stereo Temporal Difference Map” is proposed. This difference map is computed by comparing the occupancy of a cell between two consecutive frames. Based on the Stereo Temporal Difference Map information, difference fronts are extracted and are subjected to a particle based filtering mechanism. Finally, the provided dynamic features are associated to the extracted polygonal models. The result is a more compact representation of the dynamic environment.

Modeling unstructured environments with dynamic persistence grids and object delimiters in urban traffic scenarios

Modeling dynamic environments is an essential research topic in any driving assistance system. The complexity of the surrounding world, the measurement uncertainties or the unpredictable behavior of the traffic participants are the main factors that influence the detection and tracking process. In this paper we present a vision-based method for modeling and tracking unstructured dynamic environments. The proposed solution relies on raw information provided by a classified grid computed from a digital elevation map and employs two separate representation levels: a local dynamic persistence grid (DyPerGrid) that is generated as an intermediate representation level and a map of delimiters as a higher level obstacle description. A fast tracking solution is proposed by using the two models. The result is a geometrically consistent and accurate representation of the dynamic environment.

Modeling and tracking of dynamic obstacles for logistic plants using omnidirectional stereo vision

In this work we present an obstacle detection and tracking solution applied to Automated Guided Vehicles (AGVs) in industrial environments. The proposed method relies on information provided by an omnidirectional stereo vision system enabling 360 degree perception around the AGV. The stereo data is transformed into a classified digital elevation map (DEM). Based on this intermediate representation we are able to generate a set of obstacle hypotheses, each represented by a 3D cuboid and a free-form polygonal model. The cuboidal model is used for the classification of each hypothesis as “Pedestrian”, “AGV”, “Large Obstacle” or “Small Obstacle”, while the free-form polylines are used for object motion estimation relying on an Iterative Closest Point (ICP) method. The obtained measurements are subjected to a Kalman filter based tracking approach, in which the data association takes into account also the classification results.

Obstacle localization and recognition for autonomous forklifts using omnidirectional stereovision

In this paper we propose an approach for obstacle localization and recognition using omnidirectional stereovision applied to autonomous fork-lifts in industrial environments. We use omnidirectional stereovision with two fisheye cameras for the 3D perception of the surrounding environment. Using the reconstructed 3D points, a Digital Elevation Map (DEM) is constructed consisting of a 2.5D grid of elevation cells. Each cell is then classified as ground or obstacle. Further, we use the classified DEM to generate obstacle hypotheses. To ensure a higher detection rate we also propose a fast sliding window based approach relying on the monocular fisheye intensity image. The detections from both approaches are merged and are subjected to a tracking mechanism. Finally each obstacle is classified using boosting over Visual Codebook type features. The classification is refined using the classification history available from tracking. The presented approaches are integrated into a 3D visual perception system for AGVs and are of real time performance.

Tracking multiple objects in traffic scenarios using free-form obstacle delimiters and particle filters

Dynamic environment representation is an important research task in the field of advanced driving assistance systems. Usually, the tracking process is influenced by several factors, such as the unpredictable and deformable nature of the obstacles, the measurement uncertainties or the occlusions. This paper presents a stereo-vision based approach for tracking multiple objects in unstructured environments. The proposed technique relies on measurement data provided by an intermediate grid map and the object delimiters extracted from this grid. We present a particle filter based tracking solution in which a particle state is described by two components: the dynamic object parameters, and the objects geometry. In order to solve the high dimensionality state space problem a Rao-Blackwellized Particle Filter is used. The proposed method takes into consideration the stereo uncertainties and relies on a weighting mechanism based on the particle alignment error.

Vision-based tracking of multiple objects in dynamic unstructured environments using free-form obstacle delimiters

Modeling and tracking of dynamic objects is a challenging research problem in the field of driving assistance systems. Typically, the environment to be tracked is heterogeneous and unstructured. As a consequence, the tracking system must deal with measurement uncertainties, occlusions or deformable objects. In this paper we propose a real-time object tracking solution for dynamic unstructured environments. This method relies on stereo vision-based 3D information that is mapped into an intermediate digital elevation map. We apply a recursive Bayesian approach for estimating both the obstacle dynamic parameters and its geometry. In order to compute the obstacle motion we use an Iterative Closest Points-based registration technique that takes into consideration the stereo uncertainties. In our case, the object model is represented by a reference point and N delimiter landmarks. For each target we apply a Kalman filter in order to track the obstacle position and speed. In addition, the object geometry is updated by using an independent 2×2 Kalman filter for each delimiter landmark. The proposed method works in real-time and takes into consideration the stereo uncertainties.