scholarly journals Safety Design Strategies in Highly Autonomous Drive Level 2 – Lateral Control Decomposition Concept

2021 ◽  
Vol 27 (8) ◽  
pp. 811-829
Author(s):  
Svatopluk Stolfa ◽  
Jakub Stolfa ◽  
Petr Simonik ◽  
Tomas Mrovec ◽  
Tomas Harach
Keyword(s):  
Experimental Study ◽  
Autonomous Driving ◽  
Design Strategies ◽  
Lateral Control ◽  
Safe Driving ◽  
Iso 26262 ◽  
Driving Mode ◽  
Safe State ◽  
Hazard And Risk ◽  
Fusion Concept

The paper is based on an experimental study at VSB TUO Ostrava with a DEMOCAR vehicle that simulates a real car with sensor fusion concept and a vehicle gateway to send and coordinate commands to ECUs to realize and manage autonomous driving. In this experimental study of autonomous driving vehicles control, a HARA (Hazard and Risk Analysis, ISO 26262:2018) has been done on vehicle level and strategies have been defined and implemented to manage safety situations where the car lateral control shall be hand over to a driver when in HAD 2 mode. The issue is that the switching to safe state shall not be done immediately but the vehicle has to stay in safe driving mode – fail-operational up to 4 seconds until a driver can take over. The UECE and other relevant studies show that it can take up to 6 seconds if driver/operator is not in the flow (HAD 3) and up to the 2 seconds when driver is in the flow (HAD 1). The paper makes assumptions and proposals about vehicle lateral control strategy to ensure the smooth take- over of the car by driver and its impact on control software development architectures.

Cloning Safe Driving Behavior for Self-Driving Cars using Convolutional Neural Networks

2019 ◽  
Vol 12 (2) ◽  
pp. 120-127 ◽  
Author(s):  
Wael Farag
Keyword(s):  
Gradient Descent ◽  
Autonomous Driving ◽  
Driving Behavior ◽  
Training Data ◽  
Stochastic Gradient Descent ◽  
Data Set ◽  
Safe Driving ◽  
Processing Pipeline ◽  
Self Driving Cars ◽  
And Training

Background: In this paper, a Convolutional Neural Network (CNN) to learn safe driving behavior and smooth steering manoeuvring, is proposed as an empowerment of autonomous driving technologies. The training data is collected from a front-facing camera and the steering commands issued by an experienced driver driving in traffic as well as urban roads. Methods: This data is then used to train the proposed CNN to facilitate what it is called “Behavioral Cloning”. The proposed Behavior Cloning CNN is named as “BCNet”, and its deep seventeen-layer architecture has been selected after extensive trials. The BCNet got trained using Adam’s optimization algorithm as a variant of the Stochastic Gradient Descent (SGD) technique. Results: The paper goes through the development and training process in details and shows the image processing pipeline harnessed in the development. Conclusion: The proposed approach proved successful in cloning the driving behavior embedded in the training data set after extensive simulations.

scholarly journals An adaptive system for autonomous driving

2020 ◽  
Vol 28 (3) ◽  
pp. 1189-1212
Author(s):  
Martin Zimmermann ◽  
Franz Wotawa
Keyword(s):  
Adaptive Control ◽  
Automotive Industry ◽  
Environmental Conditions ◽  
Environmental Changes ◽  
Adaptive System ◽  
Control Method ◽  
Autonomous Driving ◽  
Optimal Selection ◽  
Safe State ◽  
Over Time

Abstract Having systems that can adapt themselves in case of faults or changing environmental conditions is of growing interest for industry and especially for the automotive industry considering autonomous driving. In autonomous driving, it is vital to have a system that is able to cope with faults in order to enable the system to reach a safe state. In this paper, we present an adaptive control method that can be used for this purpose. The method selects alternative actions so that given goal states can be reached, providing the availability of a certain degree of redundancy. The action selection is based on weight models that are adapted over time, capturing the success rate of certain actions. Besides the method, we present a Java implementation and its validation based on two case studies motivated by the requirements of the autonomous driving domain. We show that the presented approach is applicable both in case of environmental changes but also in case of faults occurring during operation. In the latter case, the methods provide an adaptive behavior very much close to the optimal selection.

Transfer Mechanism and Experimental Study of Tower of Liede Bridge

2011 ◽  
Vol 105-107 ◽  
pp. 1225-1234
Author(s):  
Wei Feng Wang ◽  
Xi Long Chen ◽  
Sai Ying Shi ◽  
Heng Bin Zheng
Keyword(s):  
Experimental Study ◽  
Plane Stress ◽  
Model Tests ◽  
Transfer Mechanism ◽  
Design Calculation ◽  
Safe State ◽  
Plane Stress Problem ◽  
Static Performance ◽  
Design Requirements ◽  
Under Construction

The text, drawing Liede Bridge as project background, gives a detailed analysis about the transfer mechanism of the cable tower, with the aid of analytic theory methods, FEM calculations and model tests. While these model tests and calculations show that, on one aspect, the cable tower meets design requirements and falls into a safe state under construction load and bridge load. On another, its theoretical values are in accordance with the test values generally. To illustrate that the cable tower's transmission is in line with the plane-stress problem in Elasticity Theory verifies the overall static performance in cable tower along with the theory of design calculation.

scholarly journals Importance-Aware Semantic Segmentation for Autonomous Driving System

2017 ◽  
Author(s):  
Bi-ke Chen ◽  
Chen Gong ◽  
Jian Yang
Keyword(s):  
Deep Neural Networks ◽  
Semantic Segmentation ◽  
Autonomous Driving ◽  
Learning Models ◽  
Safe Driving ◽  
Driving System ◽  
Backward Propagation ◽  
Autonomous Driving System ◽  
Propagation Rules ◽  
Different Levels

Semantic Segmentation (SS) partitions an image into several coherent semantically meaningful parts, and classifies each part into one of the pre-determined classes. In this paper, we argue that existing SS methods cannot be reliably applied to autonomous driving system as they ignore the different importance levels of distinct classes for safe-driving. For example, pedestrians in the scene are much more important than sky when driving a car, so their segmentations should be as accurate as possible. To incorporate the importance information possessed by various object classes, this paper designs an "Importance-Aware Loss" (IAL) that specifically emphasizes the critical objects for autonomous driving. IAL operates under a hierarchical structure, and the classes with different importance are located in different levels so that they are assigned distinct weights. Furthermore, we derive the forward and backward propagation rules for IAL and apply them to deep neural networks for realizing SS in intelligent driving system. The experiments on CamVid and Cityscapes datasets reveal that by employing the proposed loss function, the existing deep learning models including FCN, SegNet and ENet are able to consistently obtain the improved segmentation results on the pre-defined important classes for safe-driving.

Sign in / Sign up

Export Citation Format

Share Document