Showing papers by "Saab Automobile AB published in 2021"

Using Reinforcement Learning to Create Control Barrier Functions for Explicit Risk Mitigation in Adversarial Environments

Abstract: Air Combat is a high-risk activity carried out by trained professionals operating sophisticated equipment. During this activity, a number of trade-offs have to be made, such as the balance between risk and efficiency. A policy that minimizes risk could have very low efficiency, and one that maximizes efficiency may involve very high risk.In this study, we use Reinforcement Learning (RL) to create Control Barrier Functions (CBF) that captures the current risk, in terms of worst-case future separation between the aircraft and an enemy missile. CBFs are usually designed manually as closed-form expressions, but for a complex system such as a guided missile, this is not possible. Instead, we solve an RL problem using high fidelity simulation models to find value functions with CBF properties, that can then be used to guarantee safety in real air combat situations. We also provide a theoretical analysis of what family of RL problems result in value functions that can be used as CBFs in this way.The proposed approach allows the pilot in an air combat scenario to set the exposure level deemed acceptable and continuously monitor the risk related to his/her own safety. Given input regarding acceptable risk, the system limits the choices of the pilot to those that guarantee future satisfaction of the provided bound.

Bandpass Filter With Tunable/Switchable In-Band Interference Rejection

Abstract: This letter presents a wideband bandpass filter (BPF) with a tunable and switchable in-band notch. This allows the RF front end to pass the signal of interest uninterrupted in the absence of an in-band interferer and reject part of the passband in case a narrowband interferer exists. The BPF is based on a two-path architecture, loaded with resonators on one of the paths to deliver an absorptive notch. The resonators are tuned with varactor diodes and are deactivated when the diodes are in forward bias. The implemented proof-of-concept hardware operates in the 4.8–7.8-GHz frequency range, with an in-band notch tunable between 5 and 6.25 GHz, when activated. The transmission lines were implemented as striplines, resulting in a measured insertion loss of 0.6 dB.

Instinctual Interference-Adaptive Low-Power Receiver With Combined Feedforward and Feedback Control

Abstract: Highly adaptive, instinctively interference-tolerant radio frequency (RF) receivers are in high demand today. To achieve high-interference robustness at low average power, receivers need to be dynamically configured to operate in low-power mode in the absence of interference and a high-power interference-tolerant mode as an “instinctual” response to the blocker. In this letter, we present an interference-adaptive receiver with a control loop and on-board commercial off-the-shelf (COTS) components that adapt a 2–6-GHz low-noise amplifier (LNA) from a low-power mode (−10-dBm $P_{\text {1 dB,IN}}$ and $\approx 280$ -mW power) to high-linearity mode (1.5-dBm $P_{\text {1 dB,IN}}$ and $\approx 1.4$ -W power) where the linearity is increased by 11.5 dB ( $> 14\times $ ) with a $5\times $ increase in consumed power. With no interference, the control loop automatically brings the LNA back to the low-power mode.

Stepped Frequency Pulse Compression With Noncoherent Radar Using Deep Learning

Abstract: A deep neural network (DNN) is used for achieving subpulse resolution in noncoherent stepped frequency waveform radar. The tradeoff between high resolution and long range in radar systems is often addressed using pulse compression, allowing both long pulses and high resolution by increasing the pulse bandwidth. This typically requires a coherent radar. In this article we present a deep learning-based solution for achieving subpulse resolution with a noncoherent radar. Our results for such a system are comparable to an equivalent coherent system for signal-to-noise ratios (SNRs) greater than 10 dB. All results are based on simulated data.

Visualization and Reduction of Mutual Coupling Between Antennas Installed on a Platform

Abstract: Mutual coupling, or equivalently, the isolation between antennas, is a key parameter in antenna system design. In this work, the previously defined impedance density is generalized, and it is demonstrated how it can be used to obtain spatial information about the mutual coupling. The generalized impedance density is a real-valued scalar and it can be visualized as a three-dimensional density in space. It is shown that there is a strong connection between regions with a positive (negative) generalized impedance density and a decrease (increase) of the coupling when an absorber is placed in that region. This predictive ability is a useful feature, which is tested for three numerical cases. The results are robust to the shape of the platform, and it can be compared across frequencies. By placing absorbers based on the generalized impedance density, it is possible to reduce the required amount of absorbers needed to obtain a certain reduction in mutual coupling. The visualization results and predictions of absorber positions are compared with a Poynting vector based method. Placing absorbers based on the generalized impedance density had a larger impact on the mutual coupling, compared to the predictions with the Poynting vector based method in the investigated cases.

Neural-Network-Based Classification of Commercial Ships From Multi-Influence Passive Signatures

Abstract: Monitoring the underwater environment is important for maritime security, marine conservation, and mine countermeasures With developments in computation and artificial intelligence, it is increasingly important to measure and classify underwater ship signatures In this work, we design an artificial neural network that classifies commercial ships based on their multi-influence signature In total, 103 ship passages were included in the considered data set, with signatures recorded as the ship crossed a line of passive underwater sensors The multi-influence signature was formed by feature-level sensor fusion of the hydroacoustic signature, the underwater electric potential, and the static and alternating magnetic signatures Ships were classified according to size, or type, as broadcast on the AIS With feature-level fusion, the neural network will optimize the relationship between different types of signatures, emphasizing features with greater predictive power At the same time, weak features, even if not independently adequate for classification, can add information that improves accuracy further The developed neural network achieved a classification accuracy of 874% when classifying according to size With augmented data to balance the classes, 850% classification accuracy was achieved when classifying according to ship type This is a large improvement on the found classification accuracy when using only hydroacoustic or electromagnetic signatures This article verifies the value of feature-level sensor fusion in classification, and provides guidance on classifier design depending on the exact ship classification task

Formulation and Solution of the Multi-agent Concurrent Search and Rescue Problem

Abstract: In this paper we formulate and solve the concurrent multi-agent search and rescue problem (C-SARP), where a multi-agent system is to concurrently search an area and assist the victims found during the search. It is widely believed that a UAV-system can help saving lives by locating and assisting victims over large inaccessible areas in the initial stages after a disaster, such as an earthquake, flood, or plane crash. In such a scenario, a natural objective is to minimize the loss of lives. Therefore, two types of uncertainties needs to be taken into account, the uncertainty in position of the victims, and the uncertainty in health over time. It is rational to start looking where victims are most likely to be found, such as the reported position of a victim in a life boat with access to a radio, but it is also rational to start looking where loss of lives is most likely to occur, such as the uncertain position of victims swimming in cold water. We show that the proposed C-SARP is NP-hard, and that the two elements of search and rescue should not be decoupled, making C-SARP substantially different from previously studied multi agent problems, including coverage, multi agent travelling salesmen problems and earlier studies of decoupled search and rescue. Finally, we provide an experimental comparison between the most promising algorithms used in the literature to address similar problems, and find that the solutions to the C-SARP reproduce the trajectories recommended in search and rescue manuals for simple problems, but outperform those trajectories in terms of expected survivability for more complex scenarios.

Optimal Selection of Model Validation Experiments: Guided by Coverage

Abstract: Modeling and Simulation (M&S) is seen as a means to mitigate the difficulties associated with increased system complexity, integration, and cross-couplings effects encountered during development of aircraft subsystems. As a consequence, knowledge of model validity is necessary for taking robust and justified design decisions. This paper presents a method for using coverage metrics to formulate an optimal model validation strategy. Three fundamentally different and industrially relevant use-cases are presented. The first use-case entails the successive identification of validation settings, and the second considers the simultaneous identification of n validation settings. The latter of these two use-cases is finally expanded to incorporate a secondary model-based objective to the optimization problem in a third use-case. The approach presented is designed to be scalable and generic to models of industrially relevant complexity. As a result, selecting experiments for validation is done objectively with little required manual effort.

Early-Life Risk Factors for Breast Cancer - Prospective Follow-up in the Northern Finland Birth Cohort 1966.

Abstract: Background: While some risk factors for breast cancer have been confirmed, less is known about the role of early biological and social risk factors for breast cancer in adult life Methods: In a prospective follow-up in the Northern Finland Birth Cohort 1966 consisting of 5308 women, 120 breast cancers were reported via national registers by the end of 2018 Early risk factors were examined with univariate and multivariate analyses using Cox regression analysis The main results are reported with hazard ratios (HR) and their 95 % confidence intervals (CI) Results: In the multivariate-adjusted models, women whose mothers lived in urban areas (HR 168; 95 % CI, 113-251) during pregnancy, were low-educated (HR 240; 95 % CI, 130-445) and had been diagnosed with breast cancer (HR 197; 95 % CI, 109-358), had a higher risk for breast cancer in adult life Lower BMI at the age of 14 associated nonsignificantly with the risk of breast cancer (Mann-Whitney U test, P = 0087) No association between birth size and breast cancer risk in adult life was found Conclusions: Early-life residence and socioeconomical conditions may have an impact on developing breast cancer in women in adult life All breast cancer cases of this study were relatively young and most of them are assumed to be premenopausal Impact: This study is one of a few prospective birth cohort studies to examine early-life socioeconomic factors and breast cancer risk in adult life This study is limited due to small number of cases

Curating Datasets for Visual Runway Detection

Abstract: In Machine Learning systems, several factors impact the performance of a trained model. The most important ones include model architecture, the amount of training time, the dataset size and diversity. In the realm of safety-critical machine learning the used datasets need to reflect the environment in which the system is intended to operate, in order to minimize the generalization gap between trained and real-world inputs. Datasets should be thoroughly prepared and requirements on the properties and characteristics of the collected data need to be specified. In our work we present a case study in which generating a synthetic dataset is accomplished based on real-world flight data from the ADS-B system, containing thousands of approaches to several airports to identify real-world statistical distributions of relevant variables to vary within our dataset sampling space. We also investigate what the effects are of training a model on synthetic data to different extents, including training on translated image sets (using domain adaptation). Our results indicate airport location to be the most critical parameter to vary. We also conclude that all experiments did benefit in performance from pre-training on synthetic data rather than using only real data, however this did not hold true in general for domain adaptation-translated images.

Radar Interference Mitigation through Active Coordination

Abstract: Intelligent transportation is heavily reliant on radar, which have unique robustness under heavy rain/fog/snow and poor light conditions. With the rapid increase of the number of radars used on modern vehicles, most operating in the same frequency band, the risk of radar interference becomes an important issue. As in radio communication, interference can be mitigated through coordination. We present and evaluate two approaches for radar interference coordination, one for FMCW and one for OFDM, and highlight their challenges and opportunities.