Development Of Autonomous Vehicles Research Articles

An efficient solver of predictive control methods for motion control of autonomous vehicles with experimental verification

Real-time performance is one of the major challenges for optimal control and MPC control. In this paper, a novel general open-source C++ MPC solver based on state-of-the-art sparse QP solver OSQP is designed to solve the motion control problem of autonomous vehicles efficiently. A new linear dynamic model is proposed by replacing heading angle error with planning lateral projection speed, whose measurement is independent of the measurement of heading angle. A general local linearization method for the nonlinear tire model is proposed. Through the theory of offset-free MPC control, the steady-state value problem is solved caused by a disturbance in the time domain systematically. The algorithm in this paper is implemented in ROS, which means that the algorithm can be directly used in the prototype development of autonomous vehicles. The proposed MPC solver can solve a linear time-varying or time-invariant system, with or without disturbances. It is very efficient and can handle different kinds of hard and soft constraints. A real-time simulation platform based on Unreal Engine 4 and CarSim is designed to verify the algorithm’s effectiveness. Real-time simulations and real vehicle experiments verify the efficiency of the algorithm.

Development and Testing of Autonomous Vehicles at Scale

Development and Testing of Autonomous Vehicles at Scale

Real-Time Embedded Implementation of Improved Object Detector for Resource-Constrained Devices

Artificial intelligence (A.I.) has revolutionised a wide range of human activities, including the accelerated development of autonomous vehicles. Self-navigating delivery robots are recent trends in A.I. applications such as multitarget object detection, image classification, and segmentation to tackle sociotechnical challenges, including the development of autonomous driving vehicles, surveillance systems, intelligent transportation, and smart traffic monitoring systems. In recent years, object detection and its deployment on embedded edge devices have seen a rise in interest compared to other perception tasks. Embedded edge devices have limited computing power, which impedes the deployment of efficient detection algorithms in resource-constrained environments. To improve on-board computational latency, edge devices often sacrifice performance, creating the need for highly efficient A.I. models. This research examines existing loss metrics and their weaknesses, and proposes an improved loss metric that can address the bounding box regression problem. Enhanced metrics were implemented in an ultraefficient YOLOv5 network and tested on the targeted datasets. The latest version of the PyTorch framework was incorporated in model development. The model was further deployed using the ROS 2 framework running on NVIDIA Jetson Xavier NX, an embedded development platform, to conduct the experiment in real time.

Pandemic effects to autonomous vehicles test operations in California.

Companies developing automated driving system (ADS) technologies have spent heavily in recent years to conduct live testing of autonomous vehicles operating in real world environments to ensure their reliable and safe operations. However, the unexpected onset and ongoing resurgent effects of the Covid-19 pandemic starting in March 2020 has serve to halt, change, or delay the achievement of these new product development test objectives. This study draws on data obtained from the California automated vehicle test program to determine the extent that testing trends, test resumptions, and test environments have been affected by the pandemic. The importance of government policies to support and enable autonomous vehicles development during pandemic conditions is highlighted.

Influence of a Passenger Position Seating on Recline Seat on a Head Injury during a Frontal Crash.

Presently, most passive safety tests are performed with a precisely specified seat position and carefully seated ATD (anthropomorphic test device) dummies. Facing the development of autonomous vehicles, as well as the need for safety verification during crashes with various seat positions such research is even more urgently needed. Apart from the numerical environment, the existing testing equipment is not validated to perform such an investigation. For example, ATDs are not validated for nonstandard seatback positions, and the most accurate method of such research is volunteer tests. The study presented here was performed on a sled test rig utilizing a 50cc Hybrid III dummy according to a full factorial experiment. In addition, input factors were selected in order to verify a safe test condition for surrogate testing. The measured value was head acceleration, which was used for calculation of a head injury criterion. What was found was an optimal seat angle −117°—at which the head injury criteria had the lowest represented value. Moreover, preliminary body dynamics showed a danger of whiplash occurrence for occupants in a fully-reclined seat.

Exploring the implications of autonomous vehicles: a comprehensive review

Over the last few years, a large emphasis has been devoted to autonomous vehicles (AVs), as vehicle automation promises a large number of benefits such as: improving mobility and minimization of energy and emissions. Additionally, AVs represent a major tool in the fight against pandemics as autonomous vehicles can be used to transport people while maintaining isolation and sterilization. Thus, manufacturers are racing to introduce AVs as fast as possible. However, laws and regulations are not yet ready for this change and the legal sector is following the development of autonomous vehicles instead of taking the lead. This paper provides a comprehensive review of the previous studies in the transportation field that involve AVs with the aim of exploring the implications of AVs on the safety, public behaviour, land use, economy, society and environment, public health, and benefits of autonomous vehicles in fighting pandemics.

Changes in Electroencephalography Activity of Sensory Areas Linked to Car Sickness in Real Driving Conditions.

Car sickness is a major concern for car passengers, and with the development of autonomous vehicles, increasing numbers of car occupants are likely to be affected. Previous laboratory studies have used EEG measurements to better understand the cerebral changes linked to symptoms. However, the dynamics of motion in labs/simulators differ from those of a real car. This study sought to identify specific cerebral changes associated with the level of car sickness experienced in real driving conditions. Nine healthy volunteers participated as front passengers in a slalom session inducing lateral movements at very low frequency (0.2 Hz). They were continuously monitored via EEG recordings and subjectively rated their level of symptoms after each slalom, using a 5-point likert scale. Car-sickness symptoms evolved concomitantly with changes in theta and alpha power in the occipital and parietal areas. These changes may reflect altered sensory integration, as well as a possible influence of sleepiness mitigating symptoms.

Research on Automatic Driving Simulation Test System Based on Digital Twin

With the rapid development of 5G infrastructure, cloud computing and other technologies to enhance the rapid development of autonomous vehicles. How to quickly, safely and efficiently test and evaluate the autonomous vehicles is the only way for that to have the ability to drive on real public road. This paper build a set of automatic driving simulation test system based on digital twin technology, describing the necessary steps taken for the system design, system implementation, and testing of a autonomous vehicle. In the system the real traffic scene in the digital space re-emerged, to achieve the real vehicle running in the real environment, while mapping to the virtual environment to achieve the vehicle in the loop, making it possible to define driving test scenarios for different types of tractor and trailer combinations and providing the free test site for autonomous vehicles. At the same time, the system provides an open interface, to verify and test various autonomous algorithms. The sensor device of the tested autonomous system is flexibly configured, and the test cost is low and the efficiency is high.

A Study on Deep Learning Model Autonomous Driving Based on Big Data

Autonomous driving requires a large amount of data to improve performance, and the authors tried to solve this problem by using CARLA simulation. However, there is a suggestion that the use of CARLA simulation in research institutes and companies that research autonomous driving such as Tesla results in poor results, and that real data, not simulations, are needed in the end. Therefore, the authors propose a method to obtain real data in real-time. In order to utilize the actual data, when the sensor installed in the vehicle recognizes the dangerous situation, the embed-ded device detects and judges the danger 5-10 seconds in advance, and the acquired various dangerous situation data is sent to the iCloud(server) for retraining with new data. Over time, the learning model's performance gets better and more perfect. The deep learning model used for training is a detection model based on a convolution neural network (CNN), and a YOLO model that shows optimal detection performance and speed will be used. The authors propose a con-nectivity vehicle technology system solution, which is an important part of autonomous driv-ing, using big data-based deep learning algorithms. Connectivity autonomous driving uses big data to provide access to a high level of autonomy. A technology that provides access to a high level of autonomy is very important for vehicle development, and continuous and stable access to big data is also important for successful connectivity autonomous driving. In this study, the authors implement and extensively evaluate the system by autoware under various settings using a popular end-to-end self-driving software Autoware on NVIDIA Corporation for the development of autonomous vehicles, and evaluate the performance of Autoware on (GPUs). Autoware is a popular open-source software project that offers a complete set of autonomous driving modules that include positioning, detection, prediction, planning, and control mod-ules, which converts Windows-based sources to ROS-based sources given a source and des-tination. The authors use the popular open-source Autoware software to provide an analysis model de-sign and the development of an autonomous vehicle equipped with a system is performed. This paper proposes a UML diagram including the design of the Deep Learning Process Model Autonomous Driving Based on Big and software in it. The software is called embed-ded software. Model-based testing is a resolution for testing embedded software.

Reliability assessment of autonomous vehicles based on the safety control structure

The recent social trends and accelerated technological progress culminated in the development of autonomous vehicles (AVs). Reliability assessment for AV systems is in high demand before its market launch. In safety-critical systems (SCSs) such as AV systems, the reliability concept should be broadened to consider more safety-related issues. In this paper, reliability is defined as the probability that the system performs satisfactorily for a given period of time under stated conditions. This paper proposes a reliability assessment framework of AV, consisting of three main stages: (i) modeling the safety control structure through the Systems-Theoretic Accident Model and Processes (STAMP); (ii) mapping the control structure and functional relationships to a directed acyclic graph (DAG); and (iii) construct a Bayesian network (BN) on DAG to assess the system reliability. The fully automated (level 5) vehicle system is shown as a numeric example to illustrate how this suggested framework works. A brief discussion on involving human factors in systems to analyze lower levels of automated vehicles is also included, demonstrating the need for further research on real case studies.

Towards a Lightweight Intrusion Detection Framework for In-Vehicle Networks

With the emergence of networked devices, from the Internet of Things (IoT) nodes and cellular phones to vehicles connected to the Internet, there has been an ever-growing expansion of attack surfaces in the Internet of Vehicles (IoV). In the past decade, there has been a rapid growth in the automotive industry as network-enabled and electronic devices are now integral parts of vehicular ecosystems. These include the development of automobile technologies, namely, Connected and Autonomous Vehicles (CAV) and electric vehicles. Attacks on IoV may lead to malfunctioning of Electronic Control Unit (ECU), brakes, control steering issues, and door lock issues that can be fatal in CAV. To mitigate these risks, there is need for a lightweight model to identify attacks on vehicular systems. In this article, an efficient model of an Intrusion Detection System (IDS) is developed to detect anomalies in the vehicular system. The dataset used in this study is an In-Vehicle Network (IVN) communication protocol, i.e., Control Area Network (CAN) dataset generated in a real-time environment. The model classifies different types of attacks on vehicles into reconnaissance, Denial of Service (DoS), and fuzzing attacks. Experimentation with performance metrics of accuracy, precision, recall, and F-1 score are compared across a variety of classification models. The results demonstrate that the proposed model outperforms other classification models.

The Impact of HVAC on the Development of Autonomous and Electric Vehicle Concepts

Automation and electrification are changing vehicles and mobility. Whereas electrification is mainly changing the powertrain, automation enables the rethinking of the vehicle and its applications. The actual driving range is an important requirement for the design of automated and electric vehicles, especially if they are part of a fleet. To size the battery accordingly, not only the consumption of the powertrain has to be estimated, but also that of the auxiliary users. Heating Ventilation and Air Conditioning (HVAC) is one of the biggest auxiliary consumers. Thus, a variable HVAC model for vehicles with electric powertrain was developed to estimate the consumption depending on vehicle size and weather scenario. After integrating the model into a tool for autonomous and electric vehicle concept development, various vehicle concepts were simulated in different weather scenarios and driving cycles with the HVAC consumption considered for battery sizing. The results indicate that the battery must be resized significantly depending on the weather scenario to achieve the same driving ranges. Furthermore, the percentage of HVAC consumption is in some cases higher than that of the powertrain for urban driving cycles, due to lower average speeds. Thus, the HVAC and its energy demand should especially be considered in the development of autonomous and electric vehicles that are primarily used in cities.

Deep Learning Based Garbage Detection for Autonomous Garbage Collection Vehicles

With the rapid technological advances and the increasing human population, the need for more production has emerged and consumption has increased accordingly. As a result of this increased consumption, more garbage has been generated. Environmental pollution caused by these garbages emerges as a problem that people have to overcome both in Turkey and in the world. Many studies have been conducted to overcome this problem. Especially today, with the development of autonomous vehicles and artificial intelligence, the solutions using these technologies have increased. In this study, a new data set was created for autonomous garbage collection vehicles and a model was proposed in which these vehicles can be used. The data set was prepared with images of garbage with paper cups, which is one of the most polluting garbage, taken in different places, and images consisting of different garbage without paper cups. Paper cups were detected from these images with pre-trained Squenzenet, VGG-19 and GoogLeNet convolutional neural networks. The performance rate of the SquenzeNet, GoogLeNet and Vgg-19 networks used in the study was found as 97.77%, 96.44% 94.66%, respectively.

Traffic Risk Modelling and Analysis under Airport-Like Simulation Environment

Traffic safety plays a crucial role in the development of autonomous vehicles which attracts significant attention in the community. It is a challenge task to ensure autonomous vehicle safety under varied traffic environment interference, especially for airport-like closed-loop conditions. To that aim, we analyze autonomous vehicle safety at typical roadway conditions and traffic state constraints (e.g., car-following state at different speed distributions) by simulating the airport-like traffic conditions. The experimental results suggest that traffic collision risk is in a positive relationship with the speed difference and distance among adjacent vehicles. More specifically, the autonomous vehicle may collide with neighbors when the time to collision (TTC) indicator is lower than 4 s, and vice versa. The research findings can help both research community and practioners obtain additional information for improving traffic safety for autonomous vehicles.

An Accelerated Safety Assessment Method for Control Policies of Autonomous Vehicles in Open Environments

With the development of autonomous vehicles (AV), safety assessment for control policies becomes more and more important. Test with actual vehicles is dangerous, while high-fidelity simulators, such as Airsim, will cost plenty of time. Thus, an accelerated assessment method is badly needed in practice. As a kind of application scene for AV, open environments have fewer uncertainties but need whole-process tests for safety assessment of control policies. In this article, we propose an accelerated safety assessment method for control policies of AV in open environments. Our method can accelerate the process of safety assessment and also has a high accuracy rate. In our test of a desert road scene, our method used 1/695 of the time of the Airsim with Unreal Engine 4 (UE4), and the accuracy rate is 92%.

An embedded software architecture for the development of a cooperative autonomous vehicle

An embedded software architecture for the development of a cooperative autonomous vehicle

Parallel Hybrid 2-Opt Flower Pollination Algorithm for Real-Time UAV Trajectory Planning on GPU

Abstract. The development of autonomous Unmanned Aerial Vehicles (UAVs) is a priority to many civilian and military organizations. An essential aspect of UAV autonomy is the ability for automatic trajectory planning. In this paper, we use a parallel Flower Pollination Algorithm (FPA) to deal with the problem's complexity and compute feasible and quasi-optimal trajectories for fixed-wing UAVs in complex 3D environments, taking into account the vehicle's flight properties. The global optimization algorithm is improved with the addition of 2-opt local search providing a significant improvement. The proposed trajectory planner in implemented and parallelized on a multicore processor (CPU) using OpenMP and a Graphics Processing Unit (GPU) using CUDA resulting in a 9.6x and a 68.5x speedup respectively compared to the sequential implementation on CPU. Index Terms—Flower Pollination Algorithm, Graphics Processing Unit, Parallel Programming, Trajectory Planning, Unmanned Aerial Vehicle.

An embedded software architecture for the development of a cooperative autonomous vehicle

An embedded software architecture for the development of a cooperative autonomous vehicle

Improving the Pedestrian Detection Performance in the Absence of Rich Training Datasets: A UK Case Study

The World Health Organization estimates that well in excess of one million of lives are lost each year due to road traffic accidents. Since the human factor is the preeminent cause behind the traffic accidents, the development of reliable Advanced Driver Assistance Systems (ADASs) and Autonomous Vehicles (AVs) is seen by many as a possible solution to improve road safety. ADASs rely on the car perception system input that consists of camera(s), LIDAR and/or radar to detect pedestrians and other objects on the road. Hardware improvements as well as advances done in employing Deep Learning techniques for object detection popularized the Convolutional Neural Networks in the area of autonomous driving research and applications. However, the availability of quality and large datasets continues to be a most important contributor to the Deep Learning based model’s performance. With this in mind, this work analyses how a YOLO-based object detection architecture responded to limited data available for training and containing low-quality images. The work focused on pedestrian detection, since vulnerable road user’s safety is a major concern within AV and ADAS research communities. The proposed model was trained and tested on data gathered from Coventry, United Kingdom, city streets. The results show that the original YOLOv3 implementation reaches a 42.18% average precision (AP) and the main challenge was in detecting small objects. Network modifications were made and our final model, based on the original YOLOv3 implementation, achieved 51.6% AP. It is also demonstrated that the employed data augmentation approach is responsible for doubling the average precision of the final model.

On Optimizing Driving Patterns of Autonomous Cargo Bikes as a Function of Distance and Speed—A Psychological Study

It is predicted that the number of autonomous vehicles will increase in the forthcoming years. In this context, this research deals with autonomous micro-mobiles, specifically autonomous cargo bikes. In the development of (partially) autonomous vehicles, it is important to consider the safety perspective of vulnerable road users, such as pedestrians, already during development to enhance acceptance and widespread use. We investigated the safety parameters of speed and distance. For this purpose, videos with varying distance and speed parameters were filmed from a third-person perspective. These were then distributed via online questionnaires and evaluated by subjects regarding the subjective perceived safety. The results show that perceived safety increases with increasing passing and yielding distance and decreasing passing speed. Even at very short distances and relatively high speeds, the behavior of the cargo bike was still considered safe. These results can serve as reference values in the development of autonomous micro-mobiles. Further, the high safety perception should encourage planners and decision-makers to give these technologies a chance and boldly tackle the mobility revolution and test possible substitutes for motorized individual transport.