Autonomous driving poses unique challenges for vehicle environment perception due to the complex driving environment where the autonomous vehicle interacts with surrounding traffic participants. Due to the limited capability of any sensor perception system, it is highly desirable that an autonomous driving vehicle could use not only information from onboard sensors (say, radar/camera/lidar) but also from remote (network) information via inter-vehicle communication systems. The collaborative information from cooperative/non-cooperative remote vehicles (along with the onboard sensor data) could substantially improve the vehicle decision making process and push autonomous driving to be safer and more reliable. Inter-vehicle communication technologies are at the stage of development for market introduction, after years of research and standardization. In this paper, we setup a dedicated short range communication (DSRC) system to provide a low-latency inter-vehicle wireless communication channel. The task is to build a record linkage between the onboard sensor data and the corresponding DSRC-transmitted remote vehicle information when both sets belong to the same object, for the purpose of enhancing host vehicle environment perceiving capability and reliability. This is a typical data association problem. The challenges mainly lie in the inherent uncertain nature of the observation data and the practical issues that information often suffers from delays and drops. We propose a track-based association approach using an interacting multiple model estimator with a sequential multiple hypothesis test (denoted as IMM-SMHT) as an ubiquitous solution to handle different situations in complicated driving scenarios. To fully exploit the potential of such a system, only position information (from the DSRC channel and onboard radar system) is used for the object matching purpose-we try to use the least amount of information to achieve a high association accuracy; additional information can be used but not currently considered. We aim to provide a real world solution, and therefore, a prototype vehicle system is built with practical consideration on market availability, cost, and sensor limitations. We design meaningful use cases for creating functionality modules from a systematic point of view. The inter-vehicle information fusion system based on the track fusion approach using the IMM-SMHT is tested in real traffic on the U.S. roads and shows promising object matching performance of significant practical feasibility.

This paper investigates Recurrent Neural Networks (RNNs), particularly Dynamic Cortex Memories (DCMs), an extension of Long Short Term Memories (LSTMs) for classification of 14 different ground types based on vibration data. Also a simple regularization technique called Sequence Boundary Dropout (SBD) is introduced, which effectively enlarges the training set and improves generalization. The neural networks perform in the time domain without any explicit feature computation, while previous state-of-the-art methods extract features mainly in the frequency domain. The presented approach does not require a time window, is causal, and works just-in-time, such that a classification can be done online at each new time step. Furthermore, we show that the neural networks outperform previous methods significantly in terms of classification accuracy. Finally, we demonstrate that the networks retrained with Random Activation Preservation (RAP) can classify very early - within a fraction of a second - but robustly at the same time in a continuous recognition scenario with varying classes.

Dedicated Short Range Communication (DSRC) systems will become ubiquitous among vehicles in the near future. Because this technology enables communication between any set of DSRC-equipped vehicles, precise knowledge of these other vehicles is available to the host car. In addition to the DSRC system, onboard radars are able to provide high fidelity dynamics measurements of other objects within the sensing range. Given these two methods of measurement, environmental perception for driver assistance systems can be greatly improved, especially if the measurements are fused together. However, this is not a trivial task because of an inherent data association problem: Given the objects detected by the radar sensor, which one is truly the DSRC message sender? In this paper, we propose a system architecture to fuse DSRC and radar data. This architecture uses a reliable statistical track-to-track association algorithm in a novel way to solve this data matching problem. We present experimental results of this architecture on a system running in real traffic situations in the U. S.

Super-Resolution Image Reconstruction is known to be sensitive to errors in assumptions such as accurate sub-pixel motion estimation. Even small errors can yield a significant degradation of image quality that complicates any follow-on task such as object detection or classification. We focus on the problem of automatic quality assessment of Super-Resolution image reconstruction. We propose a bootstrap-based method that provides an objective metric quantifying reconstruction quality and thus allowing to readjust the reconstruction.

Performance (i.e., response time, throughput, resource consumption) is a key quality metric of today's applications as it heavily affects customer satisfaction. SAP strives to identify and fix performance problems before customers face them. Therefore, performance engineering methods are applied in all stages of the software lifecycle. However, especially in the development phase continuous performance evaluations can introduce a lot of overhead for developers which hinders their broad application in practice. In order to evaluate the performance of a certain software artefact (e.g. comparing two design alternatives), a developer has to run measurements that are tailored to the software artefact under test. The use of standard benchmarks would create less overhead, but the information gain is often not sufficient to answer the specific questions of developers. In this industrial paper, we present an approach that enables exhaustive, tailored performance testing with minimal effort for developers. The approach allows to define benchmark applications through a domain-specific model and realizes the transformation of those models to benchmark applications via a generic Benchmark Framework. The application of the approach in the context of the SAP Netweaver Cloud development environment demonstrated that we can efficiently identify performance problems that would not have been detected by our existing performance test infrastructure.

This work provides an approach to create a generic situation description for advanced driver assistance systems using logic reasoning on a traffic situation knowledge base. It contains multiple objects of different type such as vehicles and infrastructure elements like roads, lanes, intersections, traffic signs, traffic lights and relations among them. Logic inference is performed to check and extend the situation description and interpret the situation e. g. by reasoning about traffic rules. The capabilities of our ontological situation description approach are shown at the example of complex intersections with several roads, lanes, vehicles and different combinations of traffic signs and traffic lights. Real-time issues are discussed thereon.

The problem of distributed detection and decision fusion in Through-the-Wall Radar Imaging (TWRI) is considered. We deal with the multi-viewing case in which images corresponding to different radar locations can be collected. We present a method to adapt conventional distributed detection schemes to the scenario when no a priori knowledge about image statistics from any view is available. Further, a new scheme for estimating quality information of local detectors in a distributed detection scenario is proposed. We apply bootstrap techniques to draw inference from the radar measurements of the behind the wall scene. Simulation results as well as experimental data are used to demonstrate the performance of the proposed approach.

Outdoor robots are faced with a variety of terrain types each possessing different characteristics. To ensure a safe traversal a robot has to infer the current ground surface from sensor readings. Recent techniques generate a model which predicts the terrain class from single vibration signals disregarding the temporal coherence between consecutive measurements. In this paper, we present a novel approach in which the final classification relies on the analysis of not only one, but several recent observations. Therefore, the probabilistic framework of the Bayes filter is adopted to the problem of terrain classification. We propose an adaptive approach which automatically adjusts its parameters according to the history of observations. To demonstrate the performance of our method we further describe and compare another technique based on temporal coherence. The evaluation using data collected from our RWI ATRV-Jr robot shows that our approach is both reactive and stable enough to detect fast terrain transitions and selective misclassifications.

For safe navigation in outdoor environments, a mobile robot should be able to estimate the type of the current and forthcoming terrain. Based on this estimation, the robot can decide if the terrain is safe and may be traversed securely, or if the terrain is potentially dangerous and must be avoided or traversed carefully. This paper presents a terrain classification approach which fuses terrain predictions based on image data with predictions made by a vibration-based method. Using color images, the robot classifies terrain in front of it. When the robot later traverses the classified area, it uses vibration data to verify its former prediction. Our experiments on 14 different terrain types show that by fusing both predictions, the classification rates are significantly larger than predictions based on data from a single sensor alone.

Vision-based robot localization in outdoor environments is difficult because of changing illumination conditions. Another problem is the rough and cluttered environment which makes it hard to use visual features that are not rotation invariant. A popular method that is rotation invariant and relatively robust to changing illumination is the Scale Invariant Feature Transform (SIFT). However, due to the computationally intensive feature extraction and image matching, localization using SIFT is slow. On the other hand, techniques which use global image features are in general less robust and exact than SIFT, but are often much faster due to fast image matching. In this paper, we present a hybrid localization approach that switches between local and global image features. For most images, the hybrid approach uses fast global features. Only in difficult situations, e.g. containing strong illumination changes, the hybrid approach switches to local features. To decide which features to use for an image, we analyze the particle cloud of the particle filter that we use for position estimation. Experiments on outdoor images taken under varying illumination conditions show that the position estimates of the hybrid approach are about as exact as the estimates of SIFT alone. However, the average localization time using the hybrid approach is more than 3.5 times faster than using SIFT.

In outdoor environments, there is a variety of different types of ground surfaces. If some of them are slippery or bumpy, for example, the ground surface itself is a possible hazard for an autonomous mobile vehicle traversing the surface. Therefore, it is beneficial if the vehicle is able to estimate, which terrain it is currently traversing. Using this estimation, the vehicle can adapt its driving style to the terrain. In this paper, we present a method for terrain classification based on vibration induced in the vehicle's body. An accelerometer mounted on the vehicle measures the vibration perpendicular to the ground surface. We experimentally compare representations of the data based on the fast Fourier transform (FFT) and on the power spectral density (PSD). Additionally, we suggest a simpler and more compact representation based on features calculated from the raw data vectors and a combination of this representation with the PSD. We train and classify the data with a support vector machine (SVM). Experiments on a large real-world dataset containing seven different terrain types evaluate our approach

Trace driven simulation is extensively used in memory system evaluation. Traditional measurement equipment such as logic analyzers currently lack of the capability to record long memory access sequences (e.g. multiple seconds or even entire benchmark runs) due to their limited sampling depth, without altering system behavior. This paper presents a system, that is capable of recording long access sequences in realtime without affecting system operation. For the first time, a classification of the SPEC CPU2000 benchmark suite along main memory access criteria is provided. Furthermore the impact of shared memory graphics on system performance affecting future system simulation methodology is investigated.

We present and compare several video transmission systems with various feedback and coding strategies for wireless real-time transmission. A channel coding system based on high-memory convolutional codes (CCs) is deployed which aims at delaying the first error within a progressively coded data frame as far as possible. A low-rate feedback channel is introduced to communicate the number of error-free received bits to the transmitter or to request additional redundancy. Thus, drift problems are avoided, since video encoder and decoder predict from identical reference frames. Several video communication systems based on forward error correction (FEC) and automatic repeat request (ARQ) methods are presented and compared. Experimental results are provided. The benefits of the proposed combination of unequal error protection (UEP) or ARQ systems with progressively coded sources and network feedback are highlighted.

The trellis of a finite Abelian group code is locally (i.e., trellis section by trellis section) related to the trellis of the corresponding dual group code which allows one to express the basic operations of the a posteriori probability (APP) decoding algorithm (defined on a single trellis section of the primal trellis) in terms of the corresponding dual trellis section. Using this local approach, any algorithm employing the same type of operations as the APP algorithm can, thus, be dualized, even if the global dual code does not exist (e.g., nongroup codes represented by a group trellis). Given this, the complexity advantage of the dual approach for high-rate codes can be generalized to a broader class of APP decoding algorithms, including suboptimum algorithms approximating the true APP, which may be more attractive in practical applications due to their reduced complexity. Moreover, the local approach opens the way for mixed approaches where the operations of the APP algorithm are not exclusively performed on the primal or dual trellis. This is inevitable if the code does not possess a trellis consisting solely of group trellis sections as, e.g., for certain terminated group or ring codes. The complexity reduction offered by applying dualization is evaluated. As examples, we give a dual implementation of a suboptimum APP decoding algorithm for tailbiting convolutional codes, as well as dual implementations of APP algorithms of the sliding-window type. Moreover, we evaluate their performance for decoding usual tailbiting codes or convolutional codes, respectively, as well as their performance as component decoders in iteratively decoded parallel concatenated schemes.

Simple but meaningful models for a mobile radio channel are introduced and a channel-coding system based on high-memory rate-compatible punctured convolutional codes with an appropriate sequential decoding algorithm, the far-end error decoder (FEED), are presented. In combination with puncturing, we devise a method for unequal error protection (UEP) and error localization within a progressively coded source message without any additional error detection code. The FEED-based channel-coding system does not aim to minimize the bit or word error probability, but to delay the first error within a data frame as far as possible. This channel-coding scheme and the FEED algorithm can be used efficiently with automatic repeat request (ARQ). We present different ARQ strategies. For all forward error-correction (FEC) schemes, bounds are specified which allow the estimation of the performance and appropriate rate allocation. We briefly discuss an efficient fine granular scalable video compression scheme, the progressive texture video codec (PTVC). The proposed scheme generates an embedded bit-stream for each frame and allows reference frames to be adjusted. These source and channel-coding algorithms are used to design several video communication systems based on FEC and ARQ methods. The resulting systems are presented and compared. Performance estimations based on bounding techniques and optimized rate-allocation algorithms are derived and applied. Experimental results show the improvement potential of the proposed systems compared to standard schemes. Video communication over very low bit-rate mobile channels with varying channel conditions is thus made possible.

The performance of high-rate tail-biting convolutional codes serially concatenated with generalized partial response channels is studied. The effect of precoders on the overall performance is investigated. Extrinsic information transfer charts are used to guide the selection of appropriate tail-biting codes and precoders. Simulation results for a magnetic recording system modeled as a serial concatenation of tail-biting codes with a generalized partial response channel are presented. In particular, rate-8/9 and -16/17 short- and long-block-length tail-biting codes are studied. In the former case, hard-decision decoded interleaved Reed-Solomon (RS) codes are used as the outer-most code, whereas in the latter case the sector-size tail-biting codes replace the RS codes traditionally used in storage systems. The results indicate that high-rate tail-biting codes deliver significant performance gains when used in conjunction with a rate-1 precoder and iterative detection/decoding. The results also show that long tail-biting codes can outperform hard-decision decoding of RS codes by 2 dB at a sector error rate of approx. 10/sup -4/.

We present two new decoding algorithms that estimate the output (path in the trellis) of a Markov source observed through a discrete memoryless channel with special application to convolutional codes. In a change of paradigm, these decoders do not exclusively aim at a low average error probability, but rather try to maximize the contiguously correct decoded subpath (where a subpath begins with the first symbol) - a concept termed far end error decoding. In addition to path estimation, these decoders provide a reliability about each subpath and, thus, enable us to localize the first decoding error without spending additional redundancy. This approach is motivated by the fact that in many applications it is much more important to deliver only error free data to the source decoder rather than to achieve a low error probability. The significant performance gain possible with this new approach is demonstrated for the example of SPIHT coded images.

We present a source-channel coding system which is well suited for transmission over channels with varying channel states as common in mobile environments or broadcast scenarios. The system consists of a progressive source coder, a high-memory convolutional code (CC) with puncturing for unequal error protection (UEP) and a sequential decoding unit referred to as far end error decoder (FEED). In contrast to the conventional design goal of channel coding systems to minimize the bit error probability the presented system aims on maximizing the contiguously error-free part of the source message. Additionally, FEED allows one to localize the first error in the decoded message and only the virtually error free part is delivered to the source decoder. For image transmission gains up to 1.2 dB in average PSNR can be obtained compared to the best known system with the same channel decoder complexity. Finally, FEED allows a trade-off between receiver complexity and quality.