تسجيل الدخول
English

آلاء الشباك

الأبحاث

​​​​​Research Title :Activity Recognition for Mechanical Systems with Complex Kinematics

Authors :Ala’a Alshubbak, Daniel Görges ​

Abstract :Activity recognition is one of the major tasks in computer science and engineering that aims to recognize and understand the actions of one or more agents from a series of observations. It assists different sectors in ensuring safety in human-machine interaction. This paper investigates and tests an encoder classifier based on an LSTM model with various attention mechanism structures over a synthetic time-series motion dataset generated with the NVIDIA Isaac simulator. The simulator simulates the motion of an excavator as an example of a complex kinematic system in a construction environment, gathering data from position sensors connected to the excavator’s main joints. The results show that fewer sensors can be used for certain types of motion classification with a large accuracy range between 66.7% and 83.3%. The encoder-LSTM model with scaled dot-product attention gives the most accurate results compared to other attention mechanism types, with around 33.4% when using data from only two sensors and around 22.2% for the four main sensors. These results are in comparison to models that used other attention mechanisms.

Keywords : Attention mechanisms, Accuracy, Image processing, Kinematics, Activity recognition, Safety, Mechanical systems, Long short-term memory (LSTM)

​​Visit

​​​​​Research Title :Self-supervised Segmentation to Pose Estimation Model for Mechanical Systems with Complex Kinematics

Authors :Ala’a Alshubbak, Cyrille Ewoudi Ewoudi, Daniel Görges ​

Abstract :Self-supervised learning is one of the recent innovations in machine learning used in computer vision fields. Few attempts have so far used this approach to design and learn a skeleton for a mechanical system with complex kinematics, such as a robot manipulator. This work demonstrates novel applications of this method for pose estimation in this domain. The proposed model consists of two stages: a self-supervised stage designed to produce semantic segmentation exemplarily for an excavator, and a traditional image processing stage to convert the segmentation to a skeleton. An investigation using two types of contrastive self-supervised learning approaches shows that the DenseCL model gives a more accurate result in comparison to the SimCLR model. Accordingly, in the first stage, DenseCL is used to learn the basic structure of the object, and then K-Net associated with DeepLabV3 is used for the segmentation task. In the second stage, traditional image processing tools are applied to classify the segments and find the nearest intersection points to the boundary regions to create the skeleton. For a model based on ResNet-50, this model surpasses the previous baseline model in this field by 23.6% AP.

Keywords : Self-Supervised Learning, Semantic Segmentation, Pose Estimation, SimCLR, DenseCL, Complex Kinematic System, Excavator

​​Visit

​​​​​Research Title :Investigation of the Performance of Different Loss Function Types Within Deep Neural Anchor-Free Object Detectors

Authors : ​Ala’a Alshubbak, Daniel Görges

Abstract :In this paper, an investigation of different IoU loss functions and a spatial attention mechanism within anchor-free object detectors is presented. Two anchor-free dense predictor models are studied: FASF and FCOS models. The models are tested on two different datasets: the benchmark COCO dataset and a small dataset called OPEDD. The results show that some loss functions and using the attention mechanism outperform their original counterparts for both the huge multi-class COCO dataset and the small unity-class dataset of OPEDD. The proposed structure is tested over different backbones: ResNet-50, ResNet-101, and ResNeXt-101. The accuracy of basic models trained over the coco dataset improves by 1.3% and 1.6% mAP for the FSAF and FCOS models based on ResNet-50, respectively. On the other hand, it increases by 2.3% and 15.8% for the same models when trained on the OPEDD dataset. The effect is interpreted using a saliency map.

Keywords : Anchor-Free Object Detection, Deep Learning, ResNet, IOU Losses, Attention Mechanism, Saliency Map.

​​Visit

​​​​​Research Title :A self-supervised pose estimation approach for construction machines

Authors : Ala’a Alshubbak, Daniel Görges ​

Abstract :Pose estimation is a computer vision task used to estimate a skeleton of dynamic systems to predict future movements. Most of the research in this direction is based on a supervised learning approach which requires a massive amount of labeled datasets. In this paper, a self-supervised three-stage model based on a contrastive learning approach is introduced for estimating a skeleton of dynamic construction machines; such as excavators without using any labeled images for the first stage. The whole model structure is divided into three stages: the pre-train stage using the SimCLR contrastive approach, and two fine-tuning stages for the transfer learning and the downstream task. The model can leverage the features and learn from a huge unlabeled dataset called ACID to two small datasets generated from NVIDIA Isaac and MATLAB Simscape simulators as well as transfer the knowledge to a smaller dataset with a ratio of 3.5% from the original ACID dataset. The results show that the proposed approach can improve the accuracy of pose estimation for heavy construction machines in real images by 11% and 13% in comparison to the normal self-supervised approach with two backbones ResNet-50 and HRNet-W32, respectively.

Keywords :Self-Supervised Learning, Pose Estimation, Heavy Construction Machines, Deep Neural Network, ResNet, HRNet, Downstream Tasks, Topdown Heatmap, Simulation

​​Visit

​​​​​Research Title :Investigation of the Performance of Different Loss Function Types Within Deep Neural Anchor-Free Object Detectors

Authors :Ala’a Alshubbak, Daniel Görges ​

Abstract :In this paper, an investigation of different IoU loss functions and a spatial attention mechanism within anchor-free object detectors is presented. Two anchor-free dense predictor models are studied: FASF and FCOS models. The models are tested on two different datasets: the benchmark COCO dataset and a small dataset called OPEDD. The results show that some loss functions and using the attention mechanism outperform their original counterparts for both the huge multi-class COCO dataset and the small unity-class dataset of OPEDD. The proposed structure is tested over different backbones: ResNet-50, ResNet-101, and ResNeXt-101. The accuracy of basic models trained over the coco dataset improves by 1.3% and 1.6% mAP for the FSAF and FCOS models based on ResNet-50, respectively. On the other hand, it increases by 2.3% and 15.8% for the same models when trained on the OPEDD dataset. The effect is interpreted using a saliency map.

Keywords :Anchor-Free Object Detection, Deep Learning, ResNet, IOU Losses, Attention Mechanism, Saliency Map.

​​Visit

​​​​​Research Title :A self-supervised pose estimation approach for construction machines

Authors :Ala’a Alshubbak, Daniel Görges ​

Abstract :Pose estimation is a computer vision task used to estimate a skeleton of dynamic systems to predict future movements. Most of the research in this direction is based on a supervised learning approach which requires a massive amount of labeled datasets. In this paper, a self-supervised three-stage model based on a contrastive learning approach is introduced for estimating a skeleton of dynamic construction machines; such as excavators without using any labeled images for the first stage. The whole model structure is divided into three stages: the pre-train stage using the SimCLR contrastive approach, and two fine-tuning stages for the transfer learning and the downstream task. The model can leverage the features and learn from a huge unlabeled dataset called ACID to two small datasets generated from NVIDIA Isaac and MATLAB Simscape simulators as well as transfer the knowledge to a smaller dataset with a ratio of 3.5% from the original ACID dataset. The results show that the proposed approach can improve the accuracy of pose estimation for heavy construction machines in real images by 11% and 13% in comparison to the normal self-supervised approach with two backbones ResNet-50 and HRNet-W32, respectively.

Keywords :Self-Supervised Learning, Pose Estimation, Heavy Construction Machines, Deep Neural Network, ResNet, HRNet, Downstream Tasks, Topdown Heatmap, Simulation

​​Visit

​​​​​Research Title :Smart Factory of the Future -More efficient, more flexible, faster and more sustainable

Authors :Hashem Badra, Ala'a Alsuhbbak, Jivka Ovtcharova ​

Abstract :Smart Factory achieves an increase in flexibility, speed, efficiency and sustainability. In addition, a Smart Factory is generally roughly the optimization of the customer order process. In this article, the Smart Factory is examined in more detail with regard to its significance, requirements and components. The article deals in particular with the digital twin and virtual commissioning. The challenges facing the Smart Factory will also be briefly discussed. Furthermore, the advantages and effects of introducing the Smart Factory are explained.

Keywords :

​​Visit

​​​​​Research Title :LEGO Mindstorms NXT for elderly and visually impaired people in need: A platform

Authors :Ala'aldeen Al-Halhouli, Hala Qitouqa, Nancy Malkosh, Alaa Shubbak, Samer Al-Gharabli, Eyad Hamad ​

Abstract :This paper presents the employment of LEGO Mindstorms NXT robotics as core component of low-cost multidisciplinary platform for assisting elderly and visually impaired people. LEGO Mindstorms system offers a plug-and-play programmable robotics toolkit, incorporating construction guides, microcontrollers and sensors, all connected via a comprehensive programming language. It facilitates, without special training and at low cost, the use of such device for interpersonal communication and for handling multiple tasks required for elderly and visually impaired people in-need. The research project provides a model for larger-scale implementation, tackling the issues of creating additional functions in order to assist people in-need. The new functions were built and programmed using MATLAB through a user-friendly Graphical User Interface (GUI). Power consumption problem, besides the integration of WiFi connection has been resolved, incorporating GPS application on smart phones enhanced the guiding and tracking functions. We believe that developing and expanding the system to encompass a range of applications beyond the initial design schematics to ease conducting a limited number of pre-described protocols. However, the beneficiaries for the proposed research would be limited to elderly people who require assistance within their household as assistive-robot to facilitate a low-cost solution for a highly demanding health circumstance.

Keywords :LEGO Mindstorms NXT, elderly, visually impaired, MATLAB, healthcare

​​Visit

​​​​​Research Title :Modeling and verification of the kinematics of passenger falls on escalators

Authors : ​Lutfi Al-Sharif, Mohammad H Dado, Nadine Habash, Zeina El Rawashdeh, Ala’a Al-Shubbak

Abstract :Passenger falls on escalators are one of the major causes of accidents. Falls can either be passenger caused or escalator caused. Unplanned stops of escalators can cause passenger falls and consequential injury in the form of cuts, bruises, finger entrapment and, in certain cases, crushing leading to suffocation. In order to better understand the kinematics of passenger falls on escalators, MATLAB-based software has been developed. It simulates the effect of an acceleration signal on a rectangular block. The most important parameter of the rectangular solid block is the center of gravity ratio, which is defined relative to the axis around which the fall can take place. The model is then practically verified by subjecting an actual wooden rectangular block of a specific center of gravity ratio to an acceleration signal and taking a video of the fall. The acceleration signal is measured and fed into the model. The video is analyzed frame by frame in order to use time-lapse photography to compare the falling block with the output of the model. Good agreement has been achieved. Data on the human body is analyzed in order to arrive at an appropriate value of the center of gravity ratio for the human body. This value is further adjusted to account for variability in the human body, and expected human behavior, up to a value of 10. This result is then used to arrive at a limit on the maximum value of acceleration form the escalator braking system in order to prevent escalator-initiated passenger falls. Applying knowledge of the variability of the expected maximum acceleration from successive escalator stops is then applied to arrive at the final value of 1.16 m·s–2 as the recommended value to be used as a design and acceptance value for escalator braking system performance in order to prevent passenger falls.

Keywords :escalator, kinematics, modeling, passenger falls, center of gravity, deceleration, jerk

​​Visit


اتصل بنا

د. آلاء الشباك

آخر الأخبار

  • مسابقة الروبوفايت للروبوتات القتالية

  • التصميم الهندسي في عصر الـ BIM

  • مسابقة الروبوفايت للروبوتات القتالية

    • ​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​
    ​​​
    ​
    ASU University. All rights reserved 2020 جميع الحقوق محفوظة لجامعة العلوم التطبيقية الخاصة