https://www.mdu.se/

mdu.sePublications
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Efficient Model Design and Self-Supervised Learning for Real-Time Autonomous Vehicle Perception
Mälardalen University, Faculty of Engineering and Health Sciences, Department of Computer Science & Engineering.ORCID iD: 0000-0001-6889-5005
2026 (English)Doctoral thesis, comprehensive summary (Other academic)
Abstract [en]

Developing vehicles capable of operating without human drivers has long been a dream. Autonomous driving systems are the culmination of this dream, potentially becoming a reality if we can overcome challenges related to accuracy, response time, safety, and hardware efficiency.

Deep Neural Networks (DNNs) represent a significant advancement in data-driven perception models, revolutionizing computer vision technologies. If developed and trained properly, these models can provide accurate results within an acceptable inference time and become reliable components of autonomous driving systems.

This thesis presents our contributions to the development of accurate yet fast and hardware-efficient DNN models for safety-critical autonomous driving applications such as lane detection, stereo vision, and trajectory prediction. It particularly highlights our neural architecture search (NAS) methods to address the aforementioned challenges. Additionally, we introduce our distinct self-supervised learning method for lane detection, which is critical for training accurate models without relying on human-labeled data. This method significantly enhances the accuracy and generalization of lane detection methods across a wide range of images featuring different weather conditions, textures, and geographies.

The research presents several key contributions. Firstly, the thesis introduces a novel NAS method for 3D lane detection, named 3DLaneNAS, which provides accurate lane capturing on mountain roads and outperforms existing models in both accuracy and latency. Secondly, we propose a fast and accurate NAS method for stereo vision, named FastStereoNet, which is crucial for computing the distance of surrounding obstacles using cameras, and demonstrates superior accuracy and latency across GPUs, CPUs, and FPGAs. Thirdly, we develop a self-supervised lane detection method named CLLD that leverages strong lane priors and a cross-similarity operation to detect lanes robustly under diverse conditions. Fourthly, we introduce a NAS method for trajectory prediction using Lidar data, named TrajectoryNAS, which achieves state-of-the-art results in trajectory prediction with the lowest inference time. Finally, the thesis contributes to model compression by introducing advanced post-training quantization (PTQ)techniques, including cluster-affine transformations and output reconstruction, which significantly improve the accuracy of ultra-low-bit models while ensuring efficiency on resource-constrained hardware.

In conclusion, this thesis has successfully introduced several innovative methods tailored for autonomous driving applications, making substantial advancements in 3D lane detection, stereo vision, lane detection, trajectory prediction, and quantization. Each proposed approach demonstrates significant improvements over existing techniques in terms of accuracy, latency, generalization, and hardware efficiency across different driving conditions and platforms.

For future work, we aim to further enhance robustness and generalization by extending our methods with vision transformers and exploring hybrid NAS–quantization pipelines. Such approaches are expected to leverage deeper contextual information from visual data while further reducing computational demands, thereby improving the reliability of autonomous driving technologies in diverse real-world environments.

Place, publisher, year, edition, pages
Västerås: Mälardalens universitet, 2026.
Series
Mälardalen University Press Dissertations, ISSN 1651-4238 ; 476
Keywords [en]
Autonomous vehicles; Computer Vision; Efficient Models; Self-supervised Learning; Quantization; Neural Architecture Search
National Category
Computer Vision and Learning Systems
Research subject
Computer Science
Identifiers
URN: urn:nbn:se:mdh:diva-78999ISBN: 978-91-7485-771-9 (print)OAI: oai:DiVA.org:mdh-78999DiVA, id: diva2:2100825
Public defence
2026-11-06, Kappa, Mälardalens universitet och digitalt, Västerås, 13:30 (English)
Opponent
Supervisors
Available from: 2026-09-14 Created: 2026-09-13 Last updated: 2026-10-16Bibliographically approved
List of papers
1. 3DLaneNAS: Neural Architecture Search for Accurate and Light-Weight 3D Lane Detection
Open this publication in new window or tab >>3DLaneNAS: Neural Architecture Search for Accurate and Light-Weight 3D Lane Detection
Show others...
2022 (English)In: Artificial Neural Networks and Machine Learning – ICANN 2022: 31st International Conference on Artificial Neural Networks, Bristol, UK, September 6–9, 2022, Proceedings, Part I / [ed] Elias Pimenidis; Plamen Angelov; Chrisina Jayne; Antonios Papaleonidas; Mehmet Aydin, Springer Science and Business Media Deutschland GmbH , 2022, p. 404-415Conference paper, Published paper (Refereed)
Abstract [en]

Lane detection is one of the most fundamental tasks for autonomous driving. It plays a crucial role in the lateral control and the precise localization of autonomous vehicles. Monocular 3D lane detection methods provide state-of-the-art results for estimating the position of lanes in 3D world coordinates using only the information obtained from the front-view camera. Recent advances in Neural Architecture Search (NAS) facilitate automated optimization of various computer vision tasks. NAS can automatically optimize monocular 3D lane detection methods to enhance the extraction and combination of visual features, consequently reducing computation loads and increasing accuracy. This paper proposes 3DLaneNAS, a multi-objective method that enhances the accuracy of monocular 3D lane detection for both short- and long-distance scenarios while at the same time providing a fair amount of hardware acceleration. 3DLaneNAS utilizes a new multi-objective energy function to optimize the architecture of feature extraction and feature fusion modules simultaneously. Moreover, a transfer learning mechanism is used to improve the convergence of the search process. Experimental results reveal that 3DLaneNAS yields a minimum of 5.2% higher accuracy and ≈ 1.33 × lower latency over competing methods on the synthetic-3D-lanes dataset. Code is at https://github.com/alizoljodi/3DLaneNAS

Place, publisher, year, edition, pages
Springer Science and Business Media Deutschland GmbH, 2022
Series
Lecture Notes in Computer Science, ISSN 0302-9743 ; 13529
Keywords
3D lane detection, Autonomous vehicles, Neural architecture search, Cameras, Extraction, Feature extraction, Autonomous driving, Detection methods, Lane detection, Lateral control, Light weight, Multi objective, Neural architectures
National Category
Vehicle and Aerospace Engineering
Identifiers
urn:nbn:se:mdh:diva-60205 (URN)10.1007/978-3-031-15919-0_34 (DOI)000866210600034 ()2-s2.0-85138760578 (Scopus ID)9783031159183 (ISBN)
Conference
ICANN 2022, Bristol, UK, 6-9 September, 2022
Available from: 2022-10-12 Created: 2022-10-12 Last updated: 2026-09-13Bibliographically approved
2. Contrastive Learning for Lane Detection via cross-similarity
Open this publication in new window or tab >>Contrastive Learning for Lane Detection via cross-similarity
2024 (English)In: Pattern Recognition Letters, ISSN 0167-8655, E-ISSN 1872-7344, Vol. 185, p. 175-183Article in journal (Refereed) Published
Abstract [en]

Detecting lane markings in road scenes poses a significant challenge due to their intricate nature, which is susceptible to unfavorable conditions. While lane markings have strong shape priors, their visibility is easily compromised by varying lighting conditions, adverse weather, occlusions by other vehicles or pedestrians, road plane changes, and fading of colors over time. The detection process is further complicated by the presence of several lane shapes and natural variations, necessitating large amounts of high-quality and diverse data to train a robust lane detection model capable of handling various real-world scenarios.In this paper, we present a novel self-supervised learning method termed Contrastive Learning for Lane Detection via Cross-Similarity (CLLD) to enhance the resilience and effectiveness of lane detection models in real-world scenarios, particularly when the visibility of lane markings are compromised. CLLD introduces a novel contrastive learning (CL) method that assesses the similarity of local features within the global context of the input image. It uses the surrounding information to predict lane markings. This is achieved by integrating local feature contrastive learning with our newly proposed operation, dubbed cross-similarity.The local feature CL concentrates on extracting features from small patches, a necessity for accurately localizing lane segments. Meanwhile, cross-similarity captures global features, enabling the detection of obscured lane segments based on their surroundings. We enhance cross-similarity by randomly masking portions of input images in the process of augmentation. Extensive experiments on TuSimple and CuLane benchmark datasets demonstrate that CLLD consistently outperforms state-of-the-art contrastive learning methods, particularly in visibility-impairing conditions like shadows, while it also delivers comparable results under normal conditions. When compared to supervised learning, CLLD still excels in challenging scenarios such as shadows and crowded scenes, which are common in real-world driving.

Place, publisher, year, edition, pages
Elsevier BV, 2024
Keywords
Contrastive learning; Lane detection; Convolutional neural networks
National Category
Computer Vision and Learning Systems
Research subject
Industrial Systems
Identifiers
urn:nbn:se:mdh:diva-78993 (URN)10.1016/j.patrec.2024.08.007 (DOI)001301208800001 ()2-s2.0-85201504705 (Scopus ID)
Available from: 2026-09-13 Created: 2026-09-13 Last updated: 2026-09-16Bibliographically approved
3. FastStereoNet: A Fast Neural Architecture Search for Improving the Inference of Disparity Estimation on Resource-Limited Platforms
Open this publication in new window or tab >>FastStereoNet: A Fast Neural Architecture Search for Improving the Inference of Disparity Estimation on Resource-Limited Platforms
Show others...
2022 (English)In: IEEE Transactions on Systems, Man & Cybernetics. Systems, ISSN 2168-2216, E-ISSN 2168-2232, Vol. 52, no 8, p. 5222-5234Article in journal (Refereed) Published
Abstract [en]

Convolutional neural networks (CNNs) provide the best accuracy for disparity estimation. However, CNNs are computationally expensive, making them unfavorable for resource-limited devices with real-time constraints. Recent advances in neural architectures search (NAS) promise opportunities in automated optimization for disparity estimation. However, the main challenge of the NAS methods is the significant amount of computing time to explore a vast search space [e.g., 1.6x10(29)] and costly training candidates. To reduce the NAS computational demand, many proxy-based NAS methods have been proposed. Despite their success, most of them are designed for comparatively small-scale learning tasks. In this article, we propose a fast NAS method, called FastStereoNet, to enable resource-aware NAS within an intractably large search space. FastStereoNet automatically searches for hardware-friendly CNN architectures based on late acceptance hill climbing (LAHC), followed by simulated annealing (SA). FastStereoNet also employs a fine-tuning with a transferred weights mechanism to improve the convergence of the search process. The collection of these ideas provides competitive results in terms of search time and strikes a balance between accuracy and efficiency. Compared to the state of the art, FastStereoNet provides 5.25x reduction in search time and 44.4x reduction in model size. These benefits are attained while yielding a comparable accuracy that enables seamless deployment of disparity estimation on resource-limited devices. Finally, FastStereoNet significantly improves the perception quality of disparity estimation deployed on field-programmable gate array and Intel Neural Compute Stick 2 accelerator in a significantly less onerous manner.

Place, publisher, year, edition, pages
Institute of Electrical and Electronics Engineers (IEEE), 2022
Keywords
Disparity estimation, machine vision, neural architecture search, optimization, transfer learning
National Category
Computer Systems
Identifiers
urn:nbn:se:mdh:diva-56844 (URN)10.1109/TSMC.2021.3123136 (DOI)000732342800001 ()2-s2.0-85120087918 (Scopus ID)
Available from: 2021-12-30 Created: 2021-12-30 Last updated: 2026-09-13Bibliographically approved
4. TrajectoryNAS: A Neural Architecture Search for Trajectory Prediction
Open this publication in new window or tab >>TrajectoryNAS: A Neural Architecture Search for Trajectory Prediction
2024 (English)In: Sensors, E-ISSN 1424-8220, Vol. 24, no 17, article id 5696Article in journal (Refereed) Published
Abstract [en]

Autonomous driving systems are a rapidly evolving technology. Trajectory prediction is a critical component of autonomous driving systems that enables safe navigation by anticipating the movement of surrounding objects. Lidar point-cloud data provide a 3D view of solid objects surrounding the ego-vehicle. Hence, trajectory prediction using Lidar point-cloud data performs better than 2D RGB cameras due to providing the distance between the target object and the ego-vehicle. However, processing point-cloud data is a costly and complicated process, and state-of-the-art 3D trajectory predictions using point-cloud data suffer from slow and erroneous predictions. State-of-the-art trajectory prediction approaches suffer from handcrafted and inefficient architectures, which can lead to low accuracy and suboptimal inference times. Neural architecture search (NAS) is a method proposed to optimize neural network models by using search algorithms to redesign architectures based on their performance and runtime. This paper introduces TrajectoryNAS, a novel neural architecture search (NAS) method designed to develop an efficient and more accurate LiDAR-based trajectory prediction model for predicting the trajectories of objects surrounding the ego vehicle. TrajectoryNAS systematically optimizes the architecture of an end-to-end trajectory prediction algorithm, incorporating all stacked components that are prerequisites for trajectory prediction, including object detection and object tracking, using metaheuristic algorithms. This approach addresses the neural architecture designs in each component of trajectory prediction, considering accuracy loss and the associated overhead latency. Our method introduces a novel multi-objective energy function that integrates accuracy and efficiency metrics, enabling the creation of a model that significantly outperforms existing approaches. Through empirical studies, TrajectoryNAS demonstrates its effectiveness in enhancing the performance of autonomous driving systems, marking a significant advancement in the field. Experimental results reveal that TrajcetoryNAS yields a minimum of 4.8 higger accuracy and 1.1* lower latency over competing methods on the NuScenes dataset.

Place, publisher, year, edition, pages
MDPI, 2024
Keywords
autonomous driving, neural architecture search, trajectory prediction, 3D point cloud
National Category
Robotics and automation
Identifiers
urn:nbn:se:mdh:diva-68518 (URN)10.3390/s24175696 (DOI)001311464900001 ()39275608 (PubMedID)2-s2.0-85203872966 (Scopus ID)
Available from: 2024-09-26 Created: 2024-09-26 Last updated: 2026-09-16Bibliographically approved
5. Cat: Post-Training Quantization Error Reduction via Cluster-based Affine Transformation
Open this publication in new window or tab >>Cat: Post-Training Quantization Error Reduction via Cluster-based Affine Transformation
Show others...
2025 (English)Manuscript (preprint) (Other (popular science, discussion, etc.))
National Category
Computer Vision and Learning Systems
Identifiers
urn:nbn:se:mdh:diva-78997 (URN)
Conference
Submitted
Available from: 2026-09-13 Created: 2026-09-13 Last updated: 2026-09-16Bibliographically approved

Open Access in DiVA

fulltext(4092 kB)60 downloads
File information
File name FULLTEXT02.pdfFile size 4092 kBChecksum SHA-512
0d985a528aa3558143bb5fd3c1e99ec16eb2d3f8bc6877390eb4ca69fb6b88d68f503f9c51967dd7b7de9cd4905e7956e467f949938e9b8c3369e4019faac8d0
Type fulltextMimetype application/pdf

Authority records

Zoljodi, Ali

Search in DiVA

By author/editor
Zoljodi, Ali
By organisation
Department of Computer Science & Engineering
Computer Vision and Learning Systems

Search outside of DiVA

GoogleGoogle Scholar
Total: 60 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

isbn
urn-nbn

Altmetric score

isbn
urn-nbn
Total: 7340 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf