Daimler (Germany)

1. What Is This Book About? From Handcrafting to Automated Assembly Lines. Generative Programming. Benefits and Applicability. I. ANALYSIS AND DESIGN METHODS AND TECHNIQUES. 2. Domain Engineering. Why Is This Chapter Worth Reading? What Is Domain Engineering? Domain Analysis. Domain Design and Domain Implementation. Application Engineering. Product-Line Practices. Key Domain Engineering Concepts. Domain. Domain Scope and Scoping. Relationships between Domains. Features and Feature Models. Method Tailoring and Specialization. Survey of Domain Analysis and Domain Engineering Methods. Feature-Oriented Domain Analysis (FODA). Organization Domain Modeling (ODM). Draco. Capture. Domain Analysis and Reuse Environment (DARE). Domain-Specific Software Architecture (DSSA) Approach. Algebraic Approach. Other Approaches. Domain Engineering and Related Approaches. Historical Notes. Summary. 3. Domain Engineering and Object-Oriented Analysis and Design. Why Is This Chapter Worth Reading? OO Technology and Reuse. Solution Space. Problem Space. Relationship between Domain Engineering and Object-Oriented Analysis and Design (OOA/D) Methods. Aspects of Integrating Domain Engineering and OOA/D Methods. Horizontal versus Vertical Methods. Selected Methods. Rational Unified Process. 00ram. Reuse-Driven Software Engineering Business (RSEB). FeatuRSEB. Domain Engineering Method for Reusable Algorithmic Libraries (DEMRAL). 4. Feature Modeling. Why Is This Chapter Worth Reading? Features Revisited. Feature Modeling. Feature Models. Feature Diagrams. Other Infon-Nation Associated with Feature Diagrams in a Feature Model. Assigning Priorities to Variable Features. Availability Sites, Binding Sites, and Binding Modes. Relationship between Feature Diagrams and Other Modeling Notations and Implementation Techniques. Single Inheritance. Multiple Inheritance. Parameterized Inheritance. Static Parameterization. Dynamic Parameterization. Implementing Constraints. Tool Support for Feature Models. Frequently Asked Questions about Feature Diagrams. Feature Modeling Process. How to Find Features. Role of Variability in Modeling. 5. The Process of Generative Programming. Why Is This Chapter Worth Reading? Generative Domain Models. Main Development Steps in Generative Programming. Adapting Domain Engineering for Generative Programming. Domain-Specific Languages. DEMRAL: Example of a Domain Engineering Method for Generative Programming. Outline of DEMRAL. Domain Analysis. Domain Definition. Domain Modeling. Domain Design. Scope Domain Model for Implementation. Identify Packages. Develop Target Architectures and Identify the Implementation Components. Identify User DSLs. Identify Interactions between DSLs. Specify DSLs and Their Translation. Configuration DSLs. Expression DSLs. Domain Implementation. II. IMPLEMENTATION TECHNOLOGIES. 6. Generic Programming. Why Is This Chapter Worth Reading? What Is Generic Programming? Generic versus Generative Programming. Generic Parameters. Parametric versus Subtype Polymorphism. Genericity in Java. Bounded versus Unbounded Polymorphism. A Fresh Look at Polymorphism. Parameterized Components. Parameterized Programming. Types, Interfaces, and Specifications. Adapters. Vertical and Horizontal Parameters. Module Expressions. C++ Standard Template Library. Iterators. Freestanding Functions versus Member Functions. Generic Methodology. Historical Notes. 7. Component-Oriented Template-Based C++ Programming Techniques. Why Is This Chapter Worth Reading? Types of System Configuration. C++ Support for Dynamic Configuration. C++ Support for Static Configuration. Static Typing. Static Binding. Inlining. Templates. Parameterized Inheritance. typedefs. Member Types. Nested Classes. Prohibiting Certain Template Instantiations. Static versus Dynamic Parameterization. Wrappers Based on Parameterized Inheritance. Template Method Based on Parameterized Inheritance. Parameterizing Binding Mode. Consistent Parameterization of Multiple Components. Static Interactions between Components. Components with Influence. Components under Influence. Structured Configurations. Recursive Components. Intelligent Configuration. 8. Aspect-Oriented Decomposition and Composition. Why Is This Chapter Worth Reading? What Is Aspect-Oriented Programming? Aspect-Oriented Decomposition Approaches. Subject-Oriented Programming. Composition Filters. Demeter / Adaptive Programming. Aspect-Oriented Decomposition and Domain Engineering. How Aspects Arise. Composition Mechanisms. Requirements on Composition Mechanisms. Example: Synchronizing a Bounded Buffer. Tangled Synchronized Stack. Separating Synchronization Using Design Patterns. Separating Synchronization Using SOP. Some Problems with Design Patterns and Some Solutions. Implementing Noninvasive, Dynamic Composition in Smalltalk. Kinds of Crosscutting. How to Express Aspects in Programming Languages. Separating Synchronization Using AspectJ Cool. Implementing Dynamic Cool in Smalltalk. Implementation Technologies for Aspect-Oriented Programming. Technologies for Implementing Aspect-Specific Abstractions. Technologies for Implementing Weaving. AOP and Specialized Language Extensions. AOP and Active Libraries. Final Remarks. 9. Generators. Why Is This Chapter Worth Reading? What Are Generators? Transformational Model of Software Development. Technologies for Building Generators. Compositional versus Transformational Generators. Kinds of Transformations. Compiler Transformations. Source-to-Source Transformations. Transformation Systems. Scheduling Transformations. Existing Transformation Systems and Their Applications. Selected Approaches to Generation. Draco. GenVoca. Approaches Based on Algebraic Specifications. 10. Static Metaprogramming in C++. Why Is This Chapter Worth Reading? What Is Metaprogramming? A Quick Tour of Metaprogramming. Static Metaprogramming. C++ as a Two-Level Language. Functional Flavor of the Static Level. Class Templates as Functions. Integers and Types as Data. Symbolic Names Instead of Variables. Constant Initialization and typedef-Statements Instead of Assignment. Template Recursion Instead of Loops. Conditional Operator and Template Specialization as Conditional Constructs. Template Metaprogramming. Template Metafunctions. Metafinctions as Arguments and Return Values of Other Metafinctions. Representing Metainformation. Member Traits. Traits Classes. Traits Templates. Example: Using Template Metafunctions and Traits Templates to Implement Type Promotions. Compile-Time Lists and Trees as Nested Templates. Compile-Time Control Structures. Explicit Selection Constructs. Template Recursion as a Looping Construct. Explicit Looping Constructs. Code Generation. Simple Code Selection. Composing Templates. Generators Based on Expression Templates. Recursive Code Expansion. Explicit Loops for Generating Code. Example: Using Static Execute Loops to Test Metafunctions. Partial Evaluation in C++. Workarounds for Partial Template Specialization. Problems of Template Metaprogramming. Historical Notes. 11. Intentional Programming. Why Is This Chapter Worth Reading? What Is Intentional Programming? Technology behind IP. System Architecture. Representing Programs in IP: The Source Graph. Source Graph + Methods = Active Source. Working with the IP Programming Environment. Editing. Further Capabilities of the IP Editor. Extending the IP System with New Intentions. Advanced Topics. Questions, Methods, and a Frameworklike Organization. Source-Pattem-Based Polymorphism. Methods as Visitors. Asking Questions Synchronously and Asynchronously. Reduction. The Philosophy behind IP. Why Do We Need Extendible Programming Environments? or What Is the Problem with Fixed Programming Languages? Moving Focus from Fixed Languages to Language Features and the Emergence of an Intention Market. Intentional Programming and Component-Based Development. Frequently Asked Questions. Summary. III. APPLICATION EXAMPLES. 12. List Container. Why Is This Chapter Worth Reading? Overview. Domain Analysis. Domain Design. Implementation Components. Manual Assembly. Specifying Lists. The Generator. Extensions. 13. Bank Account. Why Is This Chapter Worth Reading? The Successful Programming Shop. Design Pattems, Frameworks, and Components. Domain Engineering and Generative Programming. Feature Modeling. Architecture Design. Implementation Components. Configurable Class Hierarchies. Designing a Domain-Specific Language. Bank Account Generator. Testing Generators and Their Products. 14. Generative Matrix Computation Library (GMCL). Why Is This Chapter Worth Reading? Why Matrix Computations? Domain Analysis. Domain Definition. Domain Modeling. Domain Design and Implementation. Matrix Type Generation. Generating Code for Matrix Expressions. Implementing the Matrix Component in IP. APPENDICES. Appendix A: Conceptual Modeling. What Are Concepts? Theories of Concepts. Basic Terminology. The Classical View. The Probabilistic View. The Exemplar View. Summary of the Three Views. Important Issues Concerning Concepts. Stability of Concepts. Concept Core. Informational Contents of Features. Feature Composition and Relationships between Features. Quality of Features. Abstraction and Generalization. Conceptual Modeling, Object-Orientation, and Software Reuse. Appendix B: Instance-Specific Extension Protocol for Smalltalk. Appendix C: Protocol for Attaching Listener Objects in Smalltalk. Appendix D: Glossary of Matrix Computation Terms. Appendix E: Metafunction for Evaluating Dependency Tables. Glossary of Generative Programming Terms. References. Index. 020130977T04062001

The interrogation of genetic markers in environmental meta-barcoding studies is currently seriously hindered by the lack of taxonomically curated reference data sets for the targeted genes. The Protist Ribosomal Reference database (PR(2), http://ssu-rrna.org/) provides a unique access to eukaryotic small sub-unit (SSU) ribosomal RNA and DNA sequences, with curated taxonomy. The database mainly consists of nuclear-encoded protistan sequences. However, metazoans, land plants, macrosporic fungi and eukaryotic organelles (mitochondrion, plastid and others) are also included because they are useful for the analysis of high-troughput sequencing data sets. Introns and putative chimeric sequences have been also carefully checked. Taxonomic assignation of sequences consists of eight unique taxonomic fields. In total, 136 866 sequences are nuclear encoded, 45 708 (36 501 mitochondrial and 9657 chloroplastic) are from organelles, the remaining being putative chimeric sequences. The website allows the users to download sequences from the entire and partial databases (including representative sequences after clustering at a given level of similarity). Different web tools also allow searches by sequence similarity. The presence of both rRNA and rDNA sequences, taking into account introns (crucial for eukaryotic sequences), a normalized eight terms ranked-taxonomy and updates of new GenBank releases were made possible by a long-term collaboration between experts in taxonomy and computer scientists.

Within an embodied cognition framework, it is argued that presence in a virtual environment (VE) develops from the construction of a spatial-functional mental model of the VE. Two cognitive processes lead to this model: the representation of bodily actions as possible actions in the VE, and the suppression of incompatible sensory input. It is hypothesized that the conscious sense of presence reflects these two components as spatial presence and involvement. This prediction was confirmed in two studies (N = 246 and N = 296) assessing self-reports of presence and immersion experiences. Additionally, judgments of “realness” were observed as a third presence component. A second-order factor analysis showed a distinction between presence, immersion, and interaction factors. Building on these results, a thirteen-item presence scale consisting of three independent components was developed and verified using confirmatory factor analyses across the two studies.

Pedestrian detection is a rapidly evolving area in computer vision with key applications in intelligent vehicles, surveillance, and advanced robotics. The objective of this paper is to provide an overview of the current state of the art from both methodological and experimental perspectives. The first part of the paper consists of a survey. We cover the main components of a pedestrian detection system and the underlying models. The second (and larger) part of the paper contains a corresponding experimental study. We consider a diverse set of state-of-the-art systems: wavelet-based AdaBoost cascade [74], HOG/linSVM [11], NN/LRF [75], and combined shape-texture detection [23]. Experiments are performed on an extensive data set captured onboard a vehicle driving through urban environment. The data set includes many thousands of training samples as well as a 27-minute test sequence involving more than 20,000 images with annotated pedestrian locations. We consider a generic evaluation setting and one specific to pedestrian detection onboard a vehicle. Results indicate a clear advantage of HOG/linSVM at higher image resolutions and lower processing speeds, and a superiority of the wavelet-based AdaBoost cascade approach at lower image resolutions and (near) real-time processing speeds. The data set (8.5 GB) is made public for benchmarking purposes.

A newly developed method is presented which allows the characterization of the electrocatalytic properties of highly dispersed electrocatalysts in a true rotating disk electrode (RDE) configuration by attaching the catalyst powder on a glossy carbon electrode via a thin Nafion film. Complete utilization and high reproducibility of both the electrode preparation and the catalyst loading could be shown via voltammetry and CO stripping voltammetry. Furthermore RDE measurements on the electro‐oxidation of hydrogen on Pt/Vulcan showed that the effect of diffusion through the Nation film can be avoided by proper electrode preparation. Therefore, the electrode kinetics for fuel cell relevant reactions under continuous flow conditions can be measured directly without mathematical modeling.

Visual understanding of complex urban street scenes is an enabling factor for a wide range of applications. Object detection has benefited enormously from large-scale datasets, especially in the context of deep learning. For semantic urban scene understanding, however, no current dataset adequately captures the complexity of real-world urban scenes. To address this, we introduce Cityscapes, a benchmark suite and large-scale dataset to train and test approaches for pixel-level and instance-level semantic labeling. Cityscapes is comprised of a large, diverse set of stereo video sequences recorded in streets from 50 different cities. 5000 of these images have high quality pixel-level annotations; 20000 additional images have coarse annotations to enable methods that leverage large volumes of weakly-labeled data. Crucially, our effort exceeds previous attempts in terms of dataset size, annotation richness, scene variability, and complexity. Our accompanying empirical study provides an in-depth analysis of the dataset characteristics, as well as a performance evaluation of several state-of-the-art approaches based on our benchmark.

Today there are several efficient algorithms that cope with the popular and computationally expensive task of association rule mining. Actually, these algorithms are more or less described on their own. In this paper we explain the fundamentals of association rule mining and moreover derive a general framework. Based on this we describe today's approaches in context by pointing out common aspects and differences. After that we thoroughly investigate their strengths and weaknesses and carry out several runtime experiments. It turns out that the runtime behavior of the algorithms is much more similar as to be expected.

In this paper, a model-based strategy for the real-time load control of parallel hybrid vehicles is presented. The aim is to develop a fuel-optimal control which is not relying on the a priori knowledge of the future driving conditions (global optimal control), but only upon the current system operation. The methodology developed is valid for those problem that are characterized by hard constraints on the state-battery state-of-charge (SOC) in this application-and by an arc cost-fuel consumption rate-which is not an explicit function of the state. A suboptimal control is found with a proper definition of a cost function to be minimized at each time instant. The "instantaneous" cost function includes the fuel energy and the electrical energy, the latter related to the state constraints. In order to weight the two forms of energy, a new definition of the equivalence factors has been derived. The strategy has been applied to the "Hyper" prototype of DaimlerChrysler, obtained from the hybridization of the Mercedes A-Class. Simulation results illustrate the potential of the proposed control in terms of fuel economy and in keeping the deviations of SOC at a low level.

Abstract-125 years after Bertha Benz completed the first overland journey in automotive history, the Mercedes Benz S-Class S 500 INTELLIGENT DRIVE followed the same route from Mannheim to Pforzheim, Germany, in fully autonomous manner. The autonomous vehicle was equipped with close-to-production sensor hardware and relied solely on vision and radar sensors in combination with accurate digital maps to obtain a comprehensive understanding of complex traffic situations. The historic Bertha Benz Memorial Route is particularly challenging for autonomous driving. The course taken by the autonomous vehicle had a length of 103 km and covered rural roads, 23 small villages and major cities (e.g. downtown Mannheim and Heidelberg). The route posed a large variety of difficult traffic scenarios including intersections with and without traffic lights, roundabouts, and narrow passages with oncoming traffic. This paper gives an overview of the autonomous vehicle and presents details on vision and radar-based perception, digital road maps and video-based self-localization, as well as motion planning in complex urban scenarios.

Routing of data in a vehicular ad hoc network is a challenging task due to the high dynamics of such a network. Recently, it was shown for the case of highway traffic that position-based routing approaches can very well deal with the high mobility of network nodes. However, baseline position-based routing has difficulties to handle two-dimensional scenarios with obstacles (buildings) and voids as it is the case for city scenarios. In this paper we analyze a position-based routing approach that makes use of the navigational systems of vehicles. By means of simulation we compare this approach with non-position-based ad hoc routing strategies (dynamic source routing and ad-hoc on-demand distance vector routing). The simulation makes use of highly realistic vehicle movement patterns derived from Daimler-Chrysler's Videlio traffic simulator. While DSR's performance is limited due to problems with scalability and handling mobility, both AODV and the position-based approach show good performances with the position-based approach outperforming AODV.

Le Mode`le de Culture Fit explique la manie`re dont l’environnement socio‐culturel influence la culture interne au travail et les pratiques de la direction des ressources humaines. Ce mode`le a e´te´ teste´ sur 2003 salarie´s d’entreprises prive´es dans 10 pays. Les participants ont rempli un questionnaire de 57 items, destine´ a` mesurer les perceptions de la direction sur 4 dimensions socio‐culturelles, 6 dimensions de culture interne au travail, et les pratiques HRM (Management des Ressources Humaines) dans 3 zones territoiriales. Une analyse ponde´re´e par re´gressions multiples, au niveau individuel, a montre´ que les directeurs qui caracte´risaient leurs environnement socio‐culturel de fac¸on fataliste, supposaient aussi que les employe´s n’e´taient pas malle´ables par nature. Ces directeurs ne pratiquaient pas l’enrichissement des postes et donnaient tout pouvoir au contrôle et a` la re´mune´ration en fonction des performances. Les directeurs qui appre´ciaient une grande loyaute´ des employe´s supposaient qu’ils remplissent entre eux des obligations re´ciproques et s’engagaient dans la voie donnant pouvoir aux pratiques HRM. Les directeurs qui percevaient le paternalisme et une forte distance de l’autorite´ dans leur environnement socio‐culturel, supposaient une re´activite´ des employe´s, et en outre ne pourvoyaient pas a` l’enrichissement des postes et a` la de´le´gation. Des mode`les spe´cifiques a` la culture qui mettent en relation ces 3 groupes de variables ainsi que les applications de ces recherches pour la psychologie industrielles trans‐culturellesont e´te´ de´battus.

Traffic jams are a fact of life for many car drivers. Every morning millions of drivers around the world sit motionless in their vehicles for long periods of time as they try to get to work, and then repeat the experience on their journeys home in the evening. The same thing often happens when they are driving to the coast for the weekend or to the airport to go on their holidays. They blame other drivers, increasing volumes of traffic and, inevitably, roadworks. So what has any of this got to do with physics?

It is shown that, in an initially homogeneous traffic flow, a region of high density and low average velocity of cars can spontaneously appear, if the density of cars in the flow exceeds some critical value. This region---a cluster of cars---can move with constant velocity in the opposite direction or in the direction of the flow, depending on the selected parameters and the initial conditions of the traffic flow. Based on numerical simulations, the kinetics of cluster formation and the shape of stationary moving clusters are found. The results presented can explain the appearance of a ``phantom traffic jam,'' which is observed in real traffic flow.

This paper presents an efficient shape-based object detection method based on Distance Transforms and describes its use for real-time vision on-board vehicles. The method uses a template hierarchy to capture the variety of object shapes; efficient hierarchies can be generated offline for given shape distributions using stochastic optimization techniques (i.e. simulated annealing). Online, matching involves a simultaneous coarse-to-fine approach over the shape hierarchy and over the transformation parameters. Very large speed-up factors are typically obtained when comparing this approach with the equivalent brute-force formulation; we have measured gains of several orders of magnitudes. We present experimental results on the real-time detection of traffic signs and pedestrians from a moving vehicle. Because of the highly time sensitive nature of these vision tasks, we also discuss some hardware-specific implementations of the proposed method as far as SIMD parallelism is concerned.

Summary: Structural damage of the human brain (perinatal damage, cerebral trauma, head injury, cerebrovascular and degenerative diseases, intracranial tumor, metabolic diseases, toxins, drug‐induced seizures) may lead to chronic epilepsy in survivors. Epidemiologic analyses show that a considerable time‐delay occurs between the exposure of the brain to injury and the appearance of seizures. Such seizures are usually partial or mixed, may develop at any age, and are difficult to treat. In rats subjected to structural damage of the brain induced by sustained convulsions triggered by systemic administration of the cholinergic agent pilocarpine, spontaneous seizures may develop after a mean latency of 14–15 days. The mean frequency of spontaneous recurrent convulsions remains constant for several months. Evolution of these convulsions proceeds through several electrographic and behavioral stages resembling kindling. Kindling may be otherwise induced in rodents by repeated systemic administration of convulsants or by repeated electrical stimulation of sensitive brain regions. These observations demonstrate that structural damage of the brain may lead to spontaneously recurrent convulsions (chronic epilepsy) in rats and that kindling may be involved in the evolution of such a condition. This finding suggests that kindling mechanisms underlie the development of epileptic foci from structural brain lesions. Such mechanisms may be involved in the etiology of some forms of epilepsy in humans. RÉSUMÉ Des lésions structurelles du cerveau humain (après souffrance périnatale, traumatisme cérébral, maladies cérébro‐vasculaires ou dégénératives, tumeurs intracrâniennes, maladies métaboliques, induction par substances toxiques ou par médicaments) peut entraîner une épilepsie chronique chez les survivals. Les analyses épidémiologiques montrent qu'il y a un long délai entre ľexposition du cerveau à la cause de la lésion e ľapparition des crises. De telles crises sont généralement partielles ou mixtes, peuvent survenir à n'importe quel âge, et sont de traitement difficile. Chez des rats, soumis à une lésion structurelle du cerveau induite par des convulsions prolongáees après administration systémique de pilocarpine, des crises spontanées peuvent survenir aprés une latence moyenne de 14 à 15 jours. La fréquence moyenne des convulsions récidivant spontanément reste constante pendant quelques mois. ľévolution de ces convulsions traverse plusieurs stades électrographiques et comportementaux qui ressemblent à ceux observés dans le kindling. Le kindling peut étre induit chez les rongeurs par ľadministration systémique répétée de produits convulsivants ou par une stimulation électrique répétée de régions sensibles du cerveau. Ces observations démontrent qu'une lésion structurelle du cerveau peut provoquer chez le rat des convulsions récidivant de façon spontanée, c'est‐à‐dire une épilepsie chronique et que le kindling peut être un mgéanisme d'un tel état. Ces données suggérent qu'un mécanisme de kindling soustend le développement de foyers épileptiques à partir de lésions cérébrales structurelles. De tels mécanismes peuvent être impliqués dans ľéliologie de certaines formes d'épilepsie humaine.

The assessment of extra-, intracellular and total body water (ECW, ICW, TBW) is important in many clinical situations. Bioimpedance spectroscopy (BIS) has advantages over dilution methods in terms of usability and reproducibility, but a careful analysis reveals systematic deviations in extremes of body composition and morbid states. Recent publications stress the need to set up and validate BIS equations in a wide variety of healthy subjects and patients with fluid imbalance. This paper presents two new equations for determination of ECW and ICW (referred to as body composition spectroscopy, BCS) based on Hanai mixture theory but corrected for body mass index (BMI). The equations were set up by means of cross validation using data of 152 subjects (120 healthy subjects, 32 dialysis patients) from three different centers. Validation was performed against bromide/deuterium dilution (NaBr, D2O) for ECW/TBW and total body potassium (TBK) for ICW. Agreement between BCS and the references (all subjects) was -0.4 +/- 1.4 L (mean +/- SD) for ECW, 0.2 +/- 2.0 L for ICW and -0.2 +/- 2.3 L for TBW. The ECW agreement between three independent reference methods (NaBr versus D2O-TBK) was -0.1 +/- 1.8 L for 74 subjects from two centers. Comparing the new BCS equations with the standard Hanai approach revealed an improvement in SEE for ICW and TBW by 0.6 L (24%) for all subjects, and by 1.2 L (48%) for 24 subjects with extreme BMIs (<20 and >30). BCS may be an appropriate method for body fluid volume determination over a wide range of body compositions in different states of health and disease.

The Functional Mockup Interface (FMI) is a tool independent standard for the exchange of dynamic models and for co-simulation. The development of FMI was initiated and organized by Daimler AG within the ITEA2 project MODELISAR. The primary goal is to support the exchange of simulation models between suppliers and OEMs even if a large variety of different tools are used. The FMI was developed in a close collaboration between simulation tool vendors and research institutes. In this article an overview about FMI is given and technical details about the solution are discussed.

Features of the ``stop-and-go'' phenomenon are found. First, the local phase transition ``free flow $\ensuremath{\rightarrow}$ synchronized flow'' is realized. Then the ``pinch effect'' in synchronized flow occurs. In the pinch region a complex sequence of narrow jams is self-formed. The following transformation of the narrow jams into wide jams determines a scale in distances between stop-and-go patterns.

The automotive industry is moving aggressively in the direction of advanced active safety. Dedicated short-range communication (DSRC) is a key enabling technology for the next generation of communication-based safety applications. One aspect of vehicular safety communication is the routine broadcast of messages among all equipped vehicles. Therefore, channel congestion control and broadcast performance improvement are of particular concern and need to be addressed in the overall protocol design. Furthermore, the explicit multichannel nature of DSRC necessitates a concurrent multichannel operational scheme for safety and non-safety applications. This article provides an overview of DSRC based vehicular safety communications and proposes a coherent set of protocols to address these requirements

Detecting people in images is key for several important application domains in computer vision. This paper presents an in-depth experimental study on pedestrian classification; multiple feature-classifier combinations are examined with respect to their ROC performance and efficiency. We investigate global versus local and adaptive versus nonadaptive features, as exemplified by PCA coefficients, Haar wavelets, and local receptive fields (LRFs). In terms of classifiers, we consider the popular Support Vector Machines (SVMs), feed-forward neural networks, and k-nearest neighbor classifier. Experiments are performed on a large data set consisting of 4,000 pedestrian and more than 25,000 nonpedestrian (labeled) images captured in outdoor urban environments. Statistically meaningful results are obtained by analyzing performance variances caused by varying training and test sets. Furthermore, we investigate how classification performance and training sample size are correlated. Sample size is adjusted by increasing the number of manually labeled training data or by employing automatic bootstrapping or cascade techniques. Our experiments show that the novel combination of SVMs with LRF features performs best. A boosted cascade of Haar wavelets can, however, reach quite competitive results, at a fraction of computational cost. The data set used in this paper is made public, establishing a benchmark for this important problem.