Times-series data augmentation and deep learning for construction equipment activity recognition
تقویت داده های سری زمانی و یادگیری عمیق برای شناخت فعالیت تجهیزات ساختمانی-2019
Automated, real-time, and reliable equipment activity recognition on construction sites can help to minimize idle time, improve operational efficiency, and reduce emissions. Previous efforts in activity recognition of construction equipment have explored different classification algorithms anm accelerometers and gyroscopes. These studies utilized pattern recognition approaches such as statistical models (e.g., hidden-Markov models); shallow neural networks (e.g., Artificial Neural Networks); and distance algorithms (e.g., K-nearest neighbor) to classify the time-series data collected from sensors mounted on the equipment. Such methods necessitate the segmentation of continuous operational data with fixed or dynamic windows to extract statistical features. This heuristic and manual feature extraction process is limited by human knowledge and can only extract human-specified shallow features. However, recent developments in deep neural networks, specifically recurrent neural network (RNN), presents new opportunities to classify sequential time-series data with recurrent lateral connections. RNN can automatically learn high-level representative features through the network instead of being manually designed, making it more suitable for complex activity recognition. However, the application of RNN requires a large training dataset which poses a practical challenge to obtain from real construction sites. Thus, this study presents a data-augmentation framework for generating synthetic time-series training data for an RNN-based deep learning network to accurately and reliably recognize equipment activities. The proposed methodology is validated by generating synthetic data from sample datasets, that were collected from two earthmoving operations in the real world. The synthetic data along with the collected data were used to train a long short-term memory (LSTM)-based RNN. The trained model was evaluated by comparing its performance with traditionally used classification algorithms for construction equipment activity recognition. The deep learning framework presented in this study outperformed the traditionally used machine learning classification algorithms for activity recognition regarding model accuracy and generalization.
Keywords: Construction equipment activity recognition | Inertial measurement unit | Deep learning | Time-series data augmentation | LSTM network | Big data analytics
Classification of foot drop gait characteristic due to lumbar radiculopathy using machine learning algorithms
طبقه بندی ویژگی راه رفتن قطره پا به دلیل رادیکولوپاتی کمری با استفاده از الگوریتم های یادگیری ماشین-2019
Background: Recently, the study of walking gait has received significant attention due to the importance of identifying disorders relating to gait patterns. Characterisation and classification of different common gait disorders such as foot drop in an effective and accurate manner can lead to improved diagnosis, prognosis assessment, and treatment. However, currently visual inspection is the main clinical method to evaluate gait disorders, which is reliant on the subjectivity of the observer, leading to inaccuracies. Research question: This study examines if it is feasible to use commercial off-the-shelf Inertial measurement unit sensors and supervised learning methods to distinguish foot drop gait disorder from the normal walking gait pattern. Method: The gait data collected from 56 adults diagnosed with foot drop due to L5 lumbar radiculopathy (with MRI verified compressive pathology), and 30 adults with normal gait during multiple walking trials on a flat surface. Machine learning algorithms were applied to the inertial sensor data to investigate the feasibility of classifying foot drop disorder. Results: The best three performing results were 88.45%, 86.87% and 86.08% accuracy derived from the Random Forest, SVM, and Naive Bayes classifiers respectively. After applying the wrapper feature selection technique, the top performance was from the Random Forest classifier with an overall accuracy of 93.18%. Significance: It is demonstrated that the combination of inertial sensors and machine learning algorithms, provides a promising and feasible solution to differentiating L5 radiculopathy related foot drop from normal walking gait patterns. The implication of this finding is to provide an objective method to help clinical decision making.
Keywords: Foot drop | Inertial measurement unit | Machine learning | Gait classification
Self calibration of the stereo vision system of the Changâ€™e-3 lunar rover based on the bundle block adjustment
کالیبراسیون از لحاظ سیستم دید استریو در جهت Changâ € ™ 3-e-3 ماهانه بر اساس تنظیم بلوک بسته-2017
The Chang’e-3 was the first lunar soft landing probe of China. It was composed of the lander and the lunar rover. The Chang’e-3 successful landed in the northwest of the Mare Imbrium in December 14, 2013. The lunar rover completed the movement, imaging and geological survey after landing. The lunar rover equipped with a stereo vision system which was made up of the Navcam system, the mast mechanism and the inertial measurement unit (IMU). The Navcam system composed of two cameras with the fixed focal length. The mast mechanism was a robot with three revolute joints. The stereo vision system was used to determine the position of the lunar rover, generate the digital elevation models (DEM) of the sur- rounding region and plan the moving paths of the lunar rover. The stereo vision system must be cali- brated before use. The control field could be built to calibrate the stereo vision system in the laboratory on the earth. However, the parameters of the stereo vision system would change after the launch, the orbital changes, the braking and the landing. Therefore, the stereo vision system should be self calibrated on the moon. An integrated self calibration method based on the bundle block adjustment is proposed in this paper. The bundle block adjustment uses each bundle of ray as the basic adjustment unit and the adjustment is implemented in the whole photogrammetric region. The stereo vision system can be self calibrated with the proposed method under the unknown lunar environment and all param- eters can be estimated simultaneously. The experiment was conducted in the ground lunar simulation field. The proposed method was compared with other methods such as the CAHVOR method, the vanish- ing point method, the Denavit-Hartenberg method, the factorization method and the weighted least- squares method. The analyzed result proved that the accuracy of the proposed method was superior to those of other methods. Finally, the proposed method was practical used to self calibrate the stereo vision system of the Chang’e-3 lunar rover on the moon.© 2017 International Society for Photogrammetry and Remote Sensing, Inc. (ISPRS). Published by ElsevierB.V. All rights reserved.
Keywords:Chang’e-3 | Stereo vision system | Bundle block adjustment | Product of exponentials