Sign recognition is an integral part of autonomous cars. Any misclassification of traffic signs can potentially lead to a multitude of disastrous consequences, ranging from a life-threatening accident to even a large-scale interruption of transportation services relying on autonomous cars. In this paper, we propose and examine security attacks against sign recognition systems for Deceiving Autonomous caRs with Toxic Signs (we call the proposed attacks DARTS). In particular, we introduce two novel methods to create these toxic signs. First, we propose Out-of-Distribution attacks, which expand the scope of adversarial examples by enabling the adversary to generate these starting from an arbitrary point in the image space compared to prior attacks which are restricted to existing training/test data (In-Distribution). Second, we present the Lenticular Printing attack, which relies on an optical phenomenon to deceive the traffic sign recognition system. We extensively evaluate the effectiveness of the proposed attacks in both virtual and real-world settings and consider both white-box and black-box threat models. Our results demonstrate that the proposed attacks are successful under both settings and threat models. We further show that Out-of-Distribution attacks can outperform In-Distribution attacks on classifiers defended using the adversarial training defense, exposing a new attack vector for these defenses.

DARTS: Deceiving Autonomous Cars with Toxic Signs

Traffic sign recognition is important for the development of driver assistance systems and fully autonomous vehicles. Even though GPS navigator systems works well for most of the time, there will always be situations when they fail. In these cases, robust vision based systems are required. Traffic signs are designed to have distinct colored fields separated by sharp boundaries. We propose to use locally segmented contours combined with an implicit star-shaped object model as prototypes for the different sign classes. The contours are described by Fourier descriptors. Matching of a query image to the sign prototype database is done by exhaustive search. This is done efficiently by using the correlation based matching scheme for Fourier descriptors and a fast cascaded matching scheme for enforcing the spatial requirements. We demonstrated on a publicly available database state of the art performance.

/pdf/using-fourier-descriptors-and-spatial-models-for-traffic-4w89gjczjp.pdf

Using fourier descriptors and spatial models for traffic sign recognition

Traffic sign detection and recognition using fully convolutional network guided proposals

Social distancing plays a pivotal role in preventing the spread of viral diseases illnesses such as COVID-19. By minimizing the close physical contact among people, we can reduce the chances of catching the virus and spreading it across the community. This two-part paper aims to provide a comprehensive survey on how emerging technologies, e.g., wireless and networking, artificial intelligence (AI) can enable, encourage, and even enforce social distancing practice. In this Part I, we provide a comprehensive background of social distancing including basic concepts, measurements, models, and propose various practical social distancing scenarios. We then discuss enabling wireless technologies which are especially effect- in social distancing, e.g., symptom prediction, detection and monitoring quarantined people, and contact tracing. The companion paper Part II surveys other emerging and related technologies, such as machine learning, computer vision, thermal, ultrasound, etc., and discusses open issues and challenges (e.g., privacy-preserving, scheduling, and incentive mechanisms) in implementing social distancing in practice.

/pdf/a-comprehensive-survey-of-enabling-and-emerging-technologies-30jjnsw3kp.pdf

A Comprehensive Survey of Enabling and Emerging Technologies for Social Distancing - Part I: Fundamentals and Enabling Technologies

Humans assimilate information from the traffic environment mainly through visual perception. Obviously, the dominant information required to conduct a vehicle can be acquired with visual sensors. However, in contrast to most other sensor principles, video signals contain relevant information in a highly indirect manner and hence visual sensing requires sophisticated machine vision and image understanding techniques. This paper provides an overview on the state of research in the field of machine vision for intelligent vehicles. The functional spectrum addressed covers the range from advanced driver assistance systems to autonomous driving. The organization of the article adopts the typical order in image processing pipelines that successively condense the rich information and vast amount of data in video sequences. Data-intensive low-level “early vision” techniques first extract features that are later grouped and further processed to obtain information of direct relevance for vehicle guidance. Recognition and classification schemes allow to identify specific objects in a traffic scene. Recently, semantic labeling techniques using convolutional neural networks have achieved impressive results in this field. High-level decisions of intelligent vehicles are often influenced by map data. The emerging role of machine vision in the mapping and localization process is illustrated at the example of autonomous driving. Scene representation methods are discussed that organize the information from all sensors and data sources and thus build the interface between perception and planning. Recently, vision benchmarks have been tailored to various tasks in traffic scene perception that provide a metric for the rich diversity of machine vision methods. Finally, the paper addresses computing architectures suited to real-time implementation. Throughout the paper, numerous specific examples and real world experiments with prototype vehicles are presented.

The Role of Machine Vision for Intelligent Vehicles

A system for real-time traffic sign detection is described in this paper. The system uses restricted Hough transform for circumferences in order to detect circular signs, and for straight lines for triangular ones. Some results obtained from a set of real road images captured under both normal and adverse weather conditions are presented as well in order to illustrate the robustness of the detection system. The average processing time is 30 ms per frame, what makes the system a good approach to work in real time conditions.

/pdf/fast-road-sign-detection-using-hough-transform-for-assisted-2ycruyr5jt.pdf

Fast road sign detection using hough transform for assisted driving of road vehicles

Sensor placement according to different performance measures obtained from the CRLB.The proposed approach is valid for different positioning technologies.Simulations are performed for an infrared positioning system used as example.The examples obtain sensor placements using two and three objectives.We show the relevance of the multiobjective optimization analyzing the Pareto fronts. This paper focuses on the application of a decision support system based on evolutionary multi-objective optimization for deploying sensors in an indoor localization system. Our methods aim to provide the human expert who works as the sensor resource manager with a full set of Pareto efficient solutions of the sensor placement problem. In our analysis, we use five scalar performance measures as objective functions derived from the covariance matrix of the estimation, namely the trace, determinant, maximum eigenvalue, ratio of maximum and minimum eigenvalues, and the uncertainty in a given direction. We run the multi-objective genetic algorithm to optimize these objectives and obtain the Pareto fronts. The paper includes a detailed explanation of every aspect of the system and an application of the proposed decision support system to an indoor infrared positioning system. Final results show the different placement alternatives according to the objectives and the trade-off between different accuracy performance measures can be clearly seen. This approach contributes to the current state-of-the art in the fact that we point out the problems of optimizing a single accuracy measure and propose using a decision support system that provides the resource manager with a full overview of the set of Pareto efficient solutions considering several accuracy metrics. Since the manager will know all the Pareto optimal solutions before deciding the final sensor placement scheme, this method provides more information than dealing with a single function of the weighted objectives. Additionally, we are able to use this system to optimize objectives obtained from fairly complex functions. On the contrary, recent works that are referenced in this paper need to simplify the localization process to obtain tractable problem formulations.

Sensor placement determination for range-difference positioning using evolutionary multi-objective optimization

In this paper, we describe a method for relating the precision in phase shift measurements to the navigation areas in which that precision is reached so that a coverage map is built, setting a signal-to-noise ratio (SNR) threshold level that depends on that phase precision. The method brings a novel approach to linking different areas in robotics and is applied to a mobile-robot (MR) local positioning system (LPS) in an intelligent space, where distances are computed from differential phase shift measurements with intensity modulation and direct detection (IMDD) infrared signals. A global model including the parameters of all the parts involved, e.g., optoelectronics, wireless channel, and instrumentation hardware, is developed. Furthermore, based on the model, an analytical expression deduced for the phase shift measurement is used to find the necessary SNR for a desired precision. A complete set of results, applying the coverage cells to a real building covering a path followed by an MR, is shown. The position of the MR can be known, with an accuracy value below 5 cm and tested in a basic rectangular locating cell with dimensions 3.0 m × 2.5 m.

Coverage-Mapping Method Based on a Hardware Model for Mobile-Robot Positioning in Intelligent Spaces

In this paper an infrared (IR) indoor local positioning system (LPS) is presented. The most relevant low level design aspects are addressed. Using sinusoidal amplitude-modulation (AM) of an infrared carrier, differential distances between a mobile emitter, the position of which is to be obtained, and fixed receivers are measured. The system may yield accuracies at the level of a few cm and addresses applications for which the increasingly available wireless technologies and smart phone sensors are not sufficient. Such applications comprise e.g., positioning mobile-robots in a manufacturing plant or positioning tools on a construction site. The proposed system works with an IR LED emitter, with a wide emitting angle, resulting in a less complex system than a laser-based one, but requiring an elaborate sensor design in order to have a sufficiently high signal-to-noise ratio (SNR) for successful demodulation. A detailed description of the basic locating cells (BLC), composed of five receivers is given as well as a study including all the blocks that comprise the system: emitter and detector devices, sensor electronics, phase measuring electronic system and hyperbolic trilateration module. All these blocks are modelled numerically and their relevant parameters are discussed with respect to their effect on the position error. The numerical analysis provides a method to evaluate the system as a whole. The choice of parameter values is a trade-off between accuracy, coverage and admissible dynamics of the mobile robot, or — equivalently — between SNR, field of view and real time response. Multipath is one of the biggest challenges for current indoor positioning systems requiring line-of-sight observations. The proposed system achieves multipath mitigation through an additional spread spectrum modulation of the sinusoidal AM signal, in analogy to the modulation of the microwave carrier with GNSS. Finally, a numerical analysis and an experiment using a prototypical BLC are summarized. They indicate that the system achieves a precision of 5 cm (2σ) for the coordinates in a fixed local coordinate frame.

Infrared local positioning system using phase differences

A new measuring architecture for phase-based infrared ranging applied to indoor positioning is presented. The motivation to develop this proposal was to reduce the critical effect of multipath interferences on the differential distances estimated by the ranging system. The multipath mitigation feature of the proposed system is based on applying a spread spectrum modulation on the emitted signal so that the line-of-sight component reaching the receivers can be discriminated from the non-line-of-sight components by applying a selective coherent demodulation. An overview of the system architecture is provided, together with the first approach to the model of the demodulation process and its validation. The conclusions obtained from the study of the proposal are used to provide preliminary results of the mitigation capabilities of the system, and to analyze the key issues involved in its implementation. The results show that the effectiveness of the proposal is strongly related with the infrared link bandwidth. Mitigation levels above 50% for multipath 4.5 meters longer than the direct path are reached if the infrared link bandwidth is higher than 200 MHz.

Ernesto Martin-Gorostiza

Papers

Fast road sign detection using hough transform for assisted driving of road vehicles

Sensor placement determination for range-difference positioning using evolutionary multi-objective optimization

Coverage-Mapping Method Based on a Hardware Model for Mobile-Robot Positioning in Intelligent Spaces

Infrared local positioning system using phase differences

Multipath mitigation for a phase-based infrared ranging system applied to indoor positioning