Distributed Accumulated State Density (DASD) filtering is an effective strategy for optimal multi-sensor fusion. However, the fusion of Accumulated State Densitys (ASDs) with unequal window lengths, as occurs in typical multi-sensor multiobject applications, has not yet been covered in the literature. This paper aims to fill that gap by exploring various approaches, which range from ASD cutting to more sophisticated ASD adaptation and ASD zero-padding. In consequence, we arrive at the optimal ASD-Information Matrix Fusion or ASD-Tracklet Fusion, respectively. Both provide the optimal and full ASD result with full-rate transmission of standard (single state) densities only, thus reducing the communication load compared to a corresponding transmission of full ASDs. These methods perform particularly well in practical scenarios where the models are not perfectly adapted. All approaches are discussed theoretically and are thoroughly evaluated in simulation.

Associating measurements with tracks is a crucial step in Multi-Object Tracking (MOT) to guarantee the safety of autonomous vehicles. To manage the exponentially growing number of track hypotheses, truncation becomes necessary. In the δ-Generalized Labeled Multi-Bernoulli (δ-GLMB) filter application, this truncation typically involves the ranked assignment problem, solved by Murty’s algorithm or the Gibbs sampling approach, both with limitations in terms of complexity or accuracy, respectively. With the motivation to improve these limitations, this paper addresses the ranked assignment problem arising from data association tasks with an approach that employs Graph Neural Networks (GNNs). The proposed Ranked Assignment Prediction Graph Neural Network (RAPNet) uses bipartite graphs to model the problem, harnessing the computational capabilities of deep learning. The conclusive evaluation compares the RAPNet with Murty’s algorithm and the Gibbs sampler, showing accuracy improvements compared to the Gibbs sampler.

The multi-sensor Labeled Multi-Bernoulli filter has the challenge of relying on the NP-hard multi-sensor update of the Generalized Labeled Multi-Bernoulli filter. This paper proposes the Fast Product Multi-Sensor Labeled Multi-Bernoulli filter, which is a filter for multi-sensor systems that solves this task by performing computationally simpler single-sensor Labeled Multi-Bernoulli filter updates based on a common prediction for each sensor. These single-sensor updates are then fused using a novel and efficient fusion strategy. Furthermore, the proposed filter is based on the Bayes parallel combination rule and can be seen as an efficient approximation of the multi-sensor Labeled Multi-Bernoulli filter. It enables full parallelization of the update step and benefits from sensor order independence compared to Iterated Corrector implementations. As a result, the robustness is increased, which is important for safety reasons, e.g., in autonomous driving. Our approach is evaluated on simulations, and the results are compared to an Iterated Corrector implementation of the Labeled Multi-Bernoulli filter.

The main challenge in random finite set-based multi-sensor multi-object tracking is the NP-hard association of the sensor measurements with the tracks. Using the Bayes parallel combination rule, we have recently proposed the Product Multi-sensor Generalized Labeled Multi-Bernoulli (PM-GLMB) filter, decomposing the multi-sensor update into simpler single-sensor problems with subsequent Bayes optimal fusion. This paper extends the filter to prior densities with Gaussian mixture spatial distributions, which is an inevitable prerequisite for the Product Multi-sensor Labeled Multi-Bernoulli (PM-LMB) filter we propose afterward. Finally, we evaluate the performance of both in a simulation study. In this context, we address the known performance problems of the PM-GLMB filter in crowded situations and show how the PM-LMB filter overcomes these.

Basis of a robust safety strategy for an automated driving function based on neural networks is a detailed description of its input domain, i.e. a description of the environment, in which the function is used. This is required to describe its functional system boundaries and to perform a comprehensive safety analysis. Moreover, it allows to tailor datasets specifically designed for safety related validation tests. Ontologies fulfill the task to gather expert knowledge and model information to enable computer aided processing, while using a notion understandable for humans. In this contribution, we propose a methodology for domain analysis to build up an ontology for perception of autonomous vehicles including characteristic features that become important when dealing with neural networks. Additionally, the method is demonstrated by the creation of a synthetic test dataset for an Euro NCAP-like use case.

Predicting future paths of vehicles or pedestrians is an essential task for automated vehicles to allow for planning the own trajectory. Using predicted paths, a planning algorithm can, e.g., react to anticipated manoeuvres of other traffic participants. For calculating risks of planned manoeuvres, it is essential that the predicted paths are generated with information about their uncertainty. Since today's state of the art trajectory prediction algorithms are based on deep neural networks (DNNs), the estimation of uncertainty is left to the neural networks as well, which usually provide no means of assessing how the uncertainty estimation works. In this paper, we present a combination of DNNs with Gaussian processes via Deep Kernel Learning (DKL), which combines the ability of DNNs to perform the prediction task with the advantage of Gaussian processes of having more interpretable probabilistic outputs. We propose and evaluate two different variants for the task of multimodal trajectory prediction using Stochastic Variational Gaussian Processes (SVGPs) and the recently proposed regression method Deep Sigma Point Processes (DSPPs), respectively. We evaluate the predictive distributions of both approaches on the publicly available Argoverse Motion Forecasting dataset and compare them to other, purely neural network based methods for uncertainty estimation.

Motion planning at urban intersections that accounts for the situation context, handles occlusions, and deals with measurement and prediction uncertainty is a major challenge on the way to urban automated driving. In this work, we address this challenge with a sampling-based optimization approach. For this, we formulate an optimal control problem that optimizes for low risk and high passenger comfort. The risk is calculated on the basis of the perception information and the respective uncertainty using a risk model. The risk model combines set-based methods and probabilistic approaches. Thus, the approach provides safety guarantees in a probabilistic sense, while for a vanishing risk, the formal safety guarantees of the set-based methods are inherited. By exploring all available behavior options, our approach solves decision making and longitudinal trajectory planning in one step. The available behavior options are provided by a formal representation of the situation context, which is also used to reduce calculation efforts. Occlusions are resolved using the external perception of infrastructure-mounted sensors. Yet, instead of merging external and ego perception with track-to-track fusion, the information is used in parallel. The motion planning scheme is validated through real-world experiments.

Automated vehicles need to not only perceive their environment, but also predict the possible future behavior of all detected traffic participants in order to safely navigate in complex scenarios and avoid critical situations, ranging from merging on highways to crossing urban intersections. Due to the availability of datasets with large numbers of recorded trajectories of traffic participants, deep learning based approaches can be used to model the behavior of road users. This paper proposes a convolutional network that operates on rasterized actor-centric images which encode the static and dynamic actor-environment. We predict multiple possible future trajectories for each traffic actor, which include position, velocity, acceleration, orientation, yaw rate and position uncertainty estimates. To make better use of the past movement of the actor, we propose to employ temporal convolutional networks (TCNs) and rely on uncertainties estimated from the previous object tracking stage. We evaluate our approach on the public "Argoverse Motion Forecasting" dataset, on which it won the first prize at the Argoverse Motion Forecasting Challenge, as presented on the NeurIPS 2019 workshop on "Machine Learning for Autonomous Driving".

In addition to vehicle control, drivers often perform secondary tasks that impede driving. Reduction of driver distraction is an important challenge for the safety of intelligent transportation systems. In this paper, a methodology for the detection and evaluation of driver distraction while performing secondary tasks is described and an appropriate hardware and a software environment is offered and studied. The system includes a model of normal driving, a subsystem for measuring the errors from the secondary tasks, and a module for total distraction evaluation. A new machine learning algorithm defines driver performance in lane keeping and speed maintenance on a specific road segment. To recognize the errors, a method is proposed, which compares normal driving parameters with ones obtained while conducting a secondary task. To evaluate distraction, an effective fuzzy logic algorithm is used. To verify the proposed approach, a case study with driver-in-the-loop experiments was carried out, in which participants performed the secondary task, namely chatting on a cell phone. The results presented in this research confirm its capability to detect and to precisely measure a level of abnormal driver performance.

Component based software development (CBSD) has been established in the development of automotive real-time embedded applications at Bosch. CBSD together with software product line (SPL) practice has improved software reuse, productivity, quality and complexity management, by raising the level of abstraction for software constructions and by sharing services. Although CBSD has contributed to the aforementioned improvement in the software development practice, the existing Bosch component model often requires software developers to take a close look at the implementation including models (e.g., ASCET-MD ) and even complex source code to understand software behavior and dependencies when reusing and adapting software components. This hinders the realization of the full benefits of CBSD, as the available information on the component level does not sufficiently describe important aspects of software behavior. This paper presents the concepts and case studies of 'signal flows' and 'mode dependent signal flows', which provide crucial software behavior information for real time embedded systems at the component level.

The method developed by T. Fujiwara et al. (1985) for efficiently computing the minimum distance of shortened Hamming codes using the weight distribution of their dual codes is extended to treat arbitrary shortened cyclic codes. Using this method implemented on a high-speed special-purpose processor, several classes of cyclic redundancy-check (CRC) codes with 24 and 32 parity bits are investigated. The CRC codes of each class are known to have the same minimum distance d/sub min.L/ in a certain range L of block lengths n, and within each class that CRC code has been determined the minimum distance of which exceeds d/sub min.L/ up to the largest block length. The d/sub min/ profiles of the resulting codes are presented and compared with the d/sub min/ profiles of recent suggestions of P. Merkey and E. C. Posner (1984), as well as with the d/sub min/ profile of the widely used 32 parity-bit standard code recommended in IEEE-802.>