Browsing by Author "Zhanabatyrova, Aziza"

Now showing 1 - 6 of 6

Automatic Map Update Using Dashcam Videos
(2023-07-01) Zhanabatyrova, Aziza; Souza Leite, Clayton; Xiao, Yu
A1 Alkuperäisartikkeli tieteellisessä aikakauslehdessä
Autonomous driving requires 3D maps that provide accurate and up-to-date information about semantic landmarks. Since cameras present wider availability and lower cost compared with laser scanners, vision-based mapping solutions, especially the ones using crowdsourced visual data, have attracted much attention from academia and industry. However, previous works have mainly focused on creating 3D point clouds, leaving automatic change detection as open issue. We propose a pipeline for initiating and updating 3D maps with dashcam videos, with a focus on automatic change detection based on comparison of metadata (e.g., the types and locations of traffic signs). To improve the performance of metadata generation, which depends on the accuracy of 3D object detection and localization, we introduce a novel deep learning-based pixel-wise 3D localization algorithm. The algorithm, trained directly with SfM (Structure from Motion) point cloud data, accurately locates objects in 3D space by estimating not only depth from monocular images but also lateral and height distances. In addition, we also propose a point clustering and thresholding algorithm to improve the robustness of the system to errors. We have performed experiments with different types of cameras, lighting, and weather conditions. The changes were detected with an average accuracy above 90%. The errors in the campus area were mainly due to traffic signs seen from a far distance to the vehicle and intended for pedestrians and cyclists only. We also conducted cause analysis of the detection and localization errors to measure the impact from the performance of the background technology in use.
Detecting and Classifying Changes in Traffic Rules using Induction Loop Data
(2024-01-22) Zhanabatyrova, Aziza; Xiao, Yu; Souza Leite, Clayton
A4 Artikkeli konferenssijulkaisussa
Up-to-date and accurate road maps with traffic rules are crucial for traffic safety and efficiency. However, detecting changes in traffic rules, particularly in road signs in an efficient manner still remains an open challenge. This paper proposes a method to detect and classify seven types of changes, relying on observable traffic flow changes like average vehicle speed. Our method employs a deep learning-based binary relevance approach, treating each type of change as a separate binary classification task. Input data comprise information from citywide induction loops, detailing congestion levels and average speed for each road, which vary with time and location. Our model outputs change type probabilities for 2.5 km 2 city regions for every 10 minutes. Unlike GPS traces of buses and taxis, which provide merely a partial view of the traffic, induction loop data offers a comprehensive, cost-effective view of traffic. However, there are three key challenges of utilizing induction loop data for change detection and classification, including sparse deployment of induction loops, high dimensionality of input data, and severe class imbalance. To address the first two challenges, we introduce novel strategies comprising dimensionality reduction, multi-dimensional sliding windows, neural network architectural choices such as residual connections and stateful recurrency, and data augmentation. Regarding the class imbalance, we apply sample weighing in the loss function. These novel design choices result in effective solutions with F1-scores exceeding 80% in experiments with simulated and real-world traffic data.
QoI-oriented incentive mechanisms for vehicle-based visual crowdsourcing
(2021-10-19) Pires Orozco, Rafael
Sähkötekniikan korkeakoulu | Master's thesis
The development of autonomous driving vehicles represents a major breakthrough towards new mobility models. The implementation of robust and secure autonomous driving services present numerous research challenges requiring the integration of a wide variety of technologies as well as the collaboration of many actors and stakeholders. In this thesis, the use of crowdsensing is investigated as a major resource for the provisioning of the information required by the autonomous vehicles. The motivation of the work derives from the increasing number of cameras and inter-vehicle communications facilities. The main contribution of the work is centered in the design of crowdsensing mechanisms capable of provisioning and ensuring the availability and quality of information required towards the implementation of Autonomous Driving (AD) applications. In particular, this work is centered around the creation of mapping information, which provides strong priors to AD localization systems. The work starts by identifying the quality of the relevant information required to properly identify relevant landmarks, such as traffic signs; as well as information enabling the localization of the vehicles such as the vehicle’s path. Once identified the shortcomings due to the lack and/or quality of information, the work introduces a crowdsourcing incentive mechanisms ensuring the availability of the information required for the deployment of secure AD applications.
Qualifying 5G SA for L4 Automated Vehicles in a Multi-PLMN Experimental Testbed
(2021-06-15) Pastor Figueroa, Giancarlo; Mutafungwa, Edward; Costa Requena, Jose; Li, Xuebing; El Marai, Oussama; Saba, Norshahida; Zhanabatyrova, Aziza; Xiao, Yu; Mustonen, Timo; Myrsky, Matthieu; Lammi, Lauri; Hamid, Umar Zakir Abdul; Boavida, Marta; Catalano, Sergio; Park, Hyunbin; Vikberg, Pyry; Lyytikäinen, Viljami
A4 Artikkeli konferenssijulkaisussa
National roaming, multi-SIM and edge computing constitute key 5G technologies for the cooperative perception and remote driving of L4 (automated) vehicles. To that end, this article reports our progress to trial these technologies at the multi-PLMN experimental 5G SA testbed of Aalto University, Finland. Overall, the objective is to qualify 5G as a core connectivity for connected, cooperative and automated mobility.
Structure from Motion-Based Mapping for Autonomous Driving: Practice and Experience
(2024-01-13) Zhanabatyrova, Aziza; Souza Leite, Clayton; Xiao, Yu
A1 Alkuperäisartikkeli tieteellisessä aikakauslehdessä
Accurate and up-to-date 3D maps, often represented as point clouds, are crucial for autonomous vehicles. Crowd-sourcing has emerged as a low-cost and scalable approach for collecting mapping data utilizing widely available dashcams and other sensing devices. However, it is still a non-trivial task to utilize crowdsourced data, such as dashcam images and video, to efficiently create or update high-quality point clouds using technologies like Structure from Motion (SfM). This study assesses and compares different image matching options available in open-source SfM software, analyzing their applicability and limitations for mapping urban scenes in different practical scenarios. Furthermore, the study analyzes the impact of various camera setups (i.e., the number of cameras and their placement) and weather conditions on the quality of the generated 3D point clouds in terms of completeness and accuracy. Based on these analyses, our study provides guidelines for creating more accurate point clouds.
Vision-based Road Construction Site Detection and Localization
(2020-12-14) Karthikeyan, Krishna
Sähkötekniikan korkeakoulu | Master's thesis
Autonomous driving is rapidly improving as companies and researchers are racing to create the perfect system. One of the uncertain environments which these systems face is when there is a construction site on the road. Road construction sites are dynamic and the components of the construction site can vary drastically from one site to another. The goal of this thesis work was to explore approaches for vision-based detection of construction sites on roads using images taken from a single camera such as a dashboard camera. This would be helpful for autonomous driving and navigation solutions once the construction work has been identified and localized for making informed decisions. Custom datasets have been created using existing NuScenes dataset as a starting point. Images containing road construction work from three cities - Boston, Helsinki and Singapore are included in the datasets. The two created datasets are targeted for image classification and object detection problems. Deep convolutional neural networks (CNN) based algorithms were tested on the datasets to classify and detect road construction sites. The final results of this work provides a proof of concept for detecting and localizing construction sites using a vision-based system with support for future expansion.