| 1. |
- Khound, Parthib, et al.
(author)
-
Performance Index Modeling from Fault Injection Analysis for an Autonomous Lane-Keeping System
- 2023
-
In: Proceeding of the 33rd European Safety and Reliability Conference. - : Research Publishing Services.
-
Conference paper (peer-reviewed)abstract
- A faulty sensor data could not only undermine the stability but also drastically compromise the safety of autonomoussystems. The reliability of the functional operation can be significantly enhanced, if any monitoring modules canevaluate the risk on the system for a particular fault in a sensor. Based on the estimated risk, the system can thenexecute the necessary safety operation. To develop a risk evaluating algorithm, the relation between the faults and theeffects should be known. Therefore, to establish such cause-and-effect relationship, this paper presents a performanceindexing method that quantifies the effects caused by given fault types with different intensities. Here, the consideredsystem is a lane keeping robot and the only sensor used for the functional operation is a red, green, and blue (RGB)camera. The lane keeping algorithm is modeled using a supervised artificial intelligence (AI) learning method. Toquantify the effects with performance indices (PIs), different faults are injected to the RBG camera. For an injectedfault type, the system’s PI is evaluated from the AI algorithm’s (open-loop) outcome and the lane keeping (closedloop) outcome. The lane keeping/closed-loop outcome is quantified from the trajectory data computed using thestrapdown inertial navigation algorithm with the measurement data from a 6D inertial measurement unit (IMU).
|
|
| 2. |
- Mohammed, Omar, et al.
(author)
-
Multilevel Artificial Intelligence Classification of Faulty Image Data for Enhancing Sensor Reliability
- 2023
-
In: Proceeding of the 33rd European Safety and Reliability Conference. - : Research Publishing Services.
-
Conference paper (peer-reviewed)abstract
- A multi-stage classification algorithm is proposed to predict the fault type and its associated intensity level of acamera input frame to enhance the reliability of a camera-based system. A fault injecting tool is used to generate thedataset required for the training. The model architecture mainly comprises three convolutions neural network (CNN)layers and three fully connected layers. The model achieves 93.8% accuracy for predicting a fault type. For the faultintensity prediction the accuracy significantly varies for each fault type but for some faults, the model achieves avery good prediction accuracy. However, for some other faults the accuracy can be remarkably low. The primaryreason for this gap is that the intensity levels of all considered faults can be described in a sufficiently quantitativeway, i.e., there is no sufficient metric available so far.
|
|
| 3. |
- Su, Peng, et al.
(author)
-
A Simulation-Aided Approach to Safety Analysis of Learning-Enabled Components in Automated Driving Systems
- 2023
-
In: Proceedings of 2023 IEEE 26th International Conference on Intelligent Transportation Systems (ITSC). - : Institute of Electrical and Electronics Engineers (IEEE).
-
Conference paper (peer-reviewed)abstract
- Artificial Intelligence (AI) techniques through Learning-Enabled Components (LEC) are widely employed in Automated Driving Systems (ADS) to support operation perception and other driving tasks relating to planning and control. Therefore, the risk management plays a critical role in assuring the operational safety of ADS. However, the probabilistic and nondeterministic nature of LEC challenges the safety analysis. Especially, the impacts of their functional faults and incompatible external conditions are often difficult to identify. To address this issue, this article presents a simulation-aided approach as follows: 1) A simulation-aided operational data generation service with the operational parameters extracted from the corresponding system models and specifications; 2) A Fault Injection (FI) serviceaimed at high-dimensional sensor data to evaluate the robustness and residual risks of LEC. 3) A Variational Bayesian (VB) method for encoding the collected operational data and supporting an effective estimation of the likelihood of operational conditions. As a case study, the paper presents the results of one experiment, where the behaviour of an Autonomous Emergency Braking(AEB) system is simulated under various weather conditions based on the CARLA driving simulator. A set of fault types of cameras, including solid occlusion, water drop, salt and pepper, are modelled and injected into the perception module of the AEB system in different weather conditions. The results indicate that our framework enables to identify the critical faults under various operational conditions. To approximate the critical faults in undefined weather, we also propose Variational Autoencoder(VAE) to encode the pixel-level data and estimate the likelihood.
|
|
| 4. |
- Su, Peng, et al.
(author)
-
Combining Self-Organizing Map with Reinforcement Learning for Multivariate Time Series Anomaly Detection
- 2023
-
In: Proceedings 2023 IEEE International Conference on Systems, Man, and Cybernetics (SMC). - : Institute of Electrical and Electronics Engineers Inc..
-
Conference paper (peer-reviewed)abstract
- Anomaly detection plays a critical role in condition monitors to support the trustworthiness of Cyber-Physical Systems (CPS). Detecting multivariate anomalous data in such systems is challenging due to the lack of a complete comprehension of anomalous behaviors and features. This paper proposes a framework to address time series multivariate anomaly detection problems by combining the Self-Organizing Map (SOM) with Deep Reinforcement Learning (DRL). By clustering the multivariate data, SOM creates an environment to enable the DRL agents interacting with the collected system operational data in terms of a tabular dataset. In this environment, Markov chains reveal the likely anomalous features to support the DRL agent exploring and exploiting the state-action space to maximize anomaly detection performance. We use a time series dataset, Skoltech Anomaly Benchmark (SKAB), to evaluate our framework. Compared with the best results by some currently applied methods, our framework improves the F1 score by 9%, from 0.67 to 0.73.
|
|
| 5. |
- Su, Peng, et al.
(author)
-
Enhancing Safety Assurance for Automated Driving Systems by Supporting Operation Simulation and Data Analysis
- 2023
-
In: Proceeding of the 33rd European Safety and Reliability Conference. - : Research Publishing Services.
-
Conference paper (peer-reviewed)abstract
- Automated Driving Systems (ADS) employ various techniques for operation perception, task planning and vehicle control. For driving on public roads, it is critical to guarantee the operational safety of such systems by attaining Minimal Risk Condition (MRC) despite unexpected environmental disruptions, human errors, functional faults and security attacks. This paper proposes a methodology to automatically identify potentially highly critical operational conditions by leveraging the design-time information in terms of vehicle architecture models and environment models. To identify the critical operating conditions, these design-time models are combined systematically with a variety of faults models for revealing the system behaviours in the presence of anomalies. The contributions of this paper are summarized as follows: 1) The design of a method for extracting related internal and external operational conditions from different system models. 2) The design of software services for identifying critical parameters and synthesizing operational data with fault injection. 3) The design for supporting operation simulation and data analysis.
|
|
| 6. |
- Su, Peng, et al.
(author)
-
Scheduling Resource to Deploy Monitors in Automated Driving Systems
- 2023
-
In: Dependable Computer Systems and Networks. - : Springer Nature. ; , s. 285-294
-
Conference paper (peer-reviewed)abstract
- Deep Neural Networks (DNN) constitute an important technology for operational perception in Automated Driving Systems (ADS). However, the trustworthiness of such DNN is one concern in the system engineering and quality management. Therefore, it is critical to monitor conditions and ensure the safety of the implementations for this advanced technology. One solution is to use Conditional Monitors (CM) to detect possible faults. However, such monitors challenge resource (e.g., data and memory) management of limited memory space in the ADS hardware. This paper proposes a resource scheme for deploying a monitor in ADS by integrating dynamic memory scheduling with Responsibility-Sensitive Safety (RSS). We use the car-following system as a case study to evaluate our scheme. YOLOv5 and KITTI datasets simulate a perception module where various monitors detect faults. We measure the time cost of conventional scheduling pipelines and our method. Compared with the conventional method, our scheme reduces 43.7% of execution time per cycle.
|
|
| 7. |
- Su, Peng, et al.
(author)
-
Using Fault Injection for the Training of Functions to Detect Soft Errors of DNNs in Automotive Vehicles
- 2022
-
In: New Advances in Dependability of Networks and Systems. - Cham : Springer. ; , s. 308-318
-
Conference paper (peer-reviewed)abstract
- Advanced functions based on Deep Neural Networks (DNN) have been widely used in automotive vehicles for the perception of operational conditions. To be able to fully exploit the potential benefits of higher levels of automated driving, the trustworthiness of such functions has to be properly ensured. This remains a challenging task for the industry as traditional approaches to system verification and validation, fault-tolerance design, become insufficient, due to the fact that many of these functions are inherently contextual and probabilistic in operation and failure. This paper presents a data centric approach to the fault characterization and data generation for the training of monitoring functions to detect soft errors of DNN functions during operation. In particular, a Fault Injection (FI) method has been developed to systematically inject both layer- and neuron-wise faults into the neural networks, including bit-flip, stuck-at, etc. The impacts of injected faults are then quantified via a probabilistic criterion based on Kullback-Leibler Divergence. We demonstrate the proposed approach based on the tests with an Alexnet.
|
|
| 8. |
- Yu, Yan Feng, et al.
(author)
-
Robust Safety Control for Automated Driving Systems with Perception Uncertainties
- 2023
-
In: Recent Developments in Model-Based and Data-Driven Methods for Advanced Control and Diagnosis. - : Springer Nature. ; , s. 321-331
-
Conference paper (peer-reviewed)abstract
- Safety assurance and trustworthiness guarantees represent some of the most challenging problems in the development of next generation automated driving and driving assistance systems. A systematic approach, with measures ranging from development-time modeling and simulation support to operation-time mechanisms for situation-awareness and adaptation, becomes necessary for tackling the problems. This paper presents a novel approach to safety control for automated driving under the condition of uncertain perception due to emergent properties in the environment or sensor faults. Based on the theory of optimal control, the safety controller is built upon a control barrier function and a model predictive control function. The effectiveness of the proposed strategy is evaluated on a simulation scenario created in the open-source autonomous driving simulator CARLA.
|
|
| 9. |
- Yuan, G., et al.
(author)
-
Graphene-based heat spreading materials for electronics packaging applications
- 2017
-
In: 2017 IMAPS Nordic Conference on Microelectronics Packaging, NordPac 2017, Goteborg, Sweden, 18-20 June 2017. - 9781538630556 ; , s. 172-174
-
Conference paper (peer-reviewed)abstract
- Graphene-based heat spreading materials, including graphene-based film (GBF) and graphene-based electrically conductive adhesive (G-CA), were applied to electronics packaging. The thermal performances of GBF and G-CA were analyzed by resistance temperature detector (RTD) and thermal infrared imager. When the chip was covered by GBF and G-CA, the temperature of hotspot could be reduced by 3.1°C, at heat flux of 580 W/cm2. To analyze the thermal performances of G-CA and GBF in 3D electronics packaging, the distribution of temperature and temperature profiles on the top surface of chip were analyzed by COMSOL. Both of GBF and G-CA could obviously reduce the temperature of hotspot on the top surface of chip, compared with that on the bare chip. With G-CA and GBF, the temperature of hotspot could be reduced by 8°C. It suggests that both of G-CA and GBF are good heat spreading materials for electronics packaging application.
|
|
