A depth-based hybrid approach for safe flight corridor generation in memoryless planning
- Authors: Nguyen, Thai , Murshed, Mamzur , Choudhury, Tanveer , Keogh, Kathleen , Kahandawa Appuhamillage, Gayan , Nguyen, Linh
- Date: 2023
- Type: Text , Journal article
- Relation: Sensors Vol. 23, no. 16 (2023), p.
- Full Text:
- Reviewed:
- Description: This paper presents a depth-based hybrid method to generate safe flight corridors for a memoryless local navigation planner. It is first proposed to use raw depth images as inputs in the learning-based object-detection engine with no requirement for map fusion. We then employ an object-detection network to directly predict the base of polyhedral safe corridors in a new raw depth image. Furthermore, we apply a verification procedure to eliminate any false predictions so that the resulting collision-free corridors are guaranteed. More importantly, the proposed mechanism helps produce separate safe corridors with minimal overlap that are suitable to be used as space boundaries for path planning. The average intersection of union (IoU) of corridors obtained by the proposed algorithm is less than 2%. To evaluate the effectiveness of our method, we incorporated it into a memoryless planner with a straight-line path-planning algorithm. We then tested the entire system in both synthetic and real-world obstacle-dense environments. The obtained results with very high success rates demonstrate that the proposed approach is highly capable of producing safe corridors for memoryless local planning. © 2023 by the authors.
Depth-based sampling and steering constraints for memoryless local planners
- Authors: Nguyen, Binh , Nguyen, Linh , Choudhury, Tanveer , Keogh, Kathleen , Murshed, Manzur
- Date: 2023
- Type: Text , Journal article
- Relation: Journal of Intelligent and Robotic Systems: Theory and Applications Vol. 109, no. 3 (2023), p.
- Full Text:
- Reviewed:
- Description: By utilizing only depth information, the paper introduces a novel two-stage planning approach that enhances computational efficiency and planning performances for memoryless local planners. First, a depth-based sampling technique is proposed to identify and eliminate a specific type of in-collision trajectories among sampled candidates. Specifically, all trajectories that have obscured endpoints are found through querying the depth values and will then be excluded from the sampled set, which can significantly reduce the computational workload required in collision checking. Subsequently, we apply a tailored local planning algorithm that employs a direction cost function and a depth-based steering mechanism to prevent the robot from being trapped in local minima. Our planning algorithm is theoretically proven to be complete in convex obstacle scenarios. To validate the effectiveness of our DEpth-based both Sampling and Steering (DESS) approaches, we conducted experiments in simulated environments where a quadrotor flew through cluttered regions with multiple various-sized obstacles. The experimental results show that DESS significantly reduces computation time in local planning compared to the uniform sampling method, resulting in the planned trajectory with a lower minimized cost. More importantly, our success rates for navigation to different destinations in testing scenarios are improved considerably compared to the fixed-yawing approach. © 2023, The Author(s).
Prediction of secant shear modulus and damping ratio for an extremely dilative silica sand based on machine learning techniques
- Authors: Baghbani, Abolfazl , Choudhury, Tanveer , Samui, Pijush , Costa, Susanga
- Date: 2023
- Type: Text , Journal article
- Relation: Soil Dynamics and Earthquake Engineering Vol. 165, no. (2023), p.
- Full Text: false
- Reviewed:
- Description: A better understanding of soil response to dynamic loads, including earthquakes, can result in safer designs that can reduce casualties. The damping ratio and shear modulus are critical parameters in soil dynamics, and several factors affect these parameters, including density and moisture content. The non-linear and multiple influences on these two parameters make their estimation difficult. Machine learning techniques are very powerful mapping tools with a remarkable capacity to perform nonlinear multivariate function approximations. In this study, to predict sand secant shear modulus and damping ratio from input variables, artificial neural networks (ANN) and classification and regression random forests (CRRF) were used as alternative estimators. The database was created using a series of simple shear tests that accurately assessed damping ratios and secant shear modulus to predict these two dynamic parameters. The input variables of the proposed predictive models included vertical stress, relative density and cyclic stress ratio, and its outputs included secant shear modulus and damping ratio. The Bayesian Regularization (BR) back-propagation ANN model produced correlation coefficient (R) and mean absolute error (MAE) values of 0.998 and 0.006, respectively, while CRRF models gave R and MAE values of 0.995 and 66.051, respectively. Additionally, sensitivity analysis of artificial intelligence (AI) models demonstrated that vertical stress and relative density played a vital role in predicting damping ratio, while all three parameters were important in predicting secant shear modulus. In this study, two developed artificial intelligence models were compared with existing literature models. According to the results, for test database, the existing models were able to predict the shear modulus and damping ratio with R of 0.911 and 0.918, respectively. However, the proposed ANN and CRRF models were able to predict shear modulus with R of 0.993 and 0.996, and damping ratios with R of 0.992 and 0.990, respectively. The results showed that ANNs and CRRFs were more robust than existing models for predicting damping ratio and shear modulus, as well as identifying the influence of input variables on sand dynamic properties. © 2022 Elsevier Ltd
Agoraphilic navigation algorithm in dynamic environment with obstacles motion tracking and prediction
- Authors: Hewawasam, Hasitha , Ibrahim, Yousef , Kahandawa, Gayan , Choudhury, Tanveer
- Date: 2022
- Type: Text , Journal article
- Relation: Robotica Vol. 40, no. 2 (2022), p. 329-347
- Full Text: false
- Reviewed:
- Description: This paper presents a new algorithm to navigate robots in dynamically cluttered environments. The proposed algorithm uses basic concepts of space attraction (hence the term Agoraphilic) to navigate robots through dynamic obstacles. The new algorithm in this paper is an advanced development of the original Agoraphilic navigation algorithm that was only able to navigate robots in static environments. The Agoraphilic algorithm does not look for obstacles (problems) to avoid but rather for a free space (solutions) to follow. Therefore, it is also described as an optimistic navigation algorithm. This algorithm uses only one attractive force created by the available free space. The free-space concept allows the Agoraphilic algorithm to overcome inherited challenges of general navigation algorithms. However, the original Agoraphilic algorithm has the limitation in navigating robots only in static, not in dynamic environments. The presented algorithm was developed to address this limitation of the original Agoraphilic algorithm. The new algorithm uses a developed object tracking module to identify the time-varying free spaces by tracking moving obstacles. The capacity of the algorithm was further strengthened by the new prediction module. Future space prediction allowed the algorithm to make decisions considering future growing/diminishing free spaces. This paper also includes a bench-marking study of the new algorithm compared with a recently published APF-based algorithm under a similar operating environment. Furthermore, the algorithm was validated based on experimental tests and simulation tests. © 2022 Cambridge University Press. All rights reserved.
Agoraphilic navigation algorithm under dynamic environment
- Authors: Hewawasam, Hasitha , Ibrahim, Yousef , Appuhamillage, Gayan , Choudhury, Tanveer
- Date: 2022
- Type: Text , Journal article
- Relation: IEEE/ASME Transactions on Mechatronics Vol. 27, no. 3 (2022), p. 1727-1737
- Full Text: false
- Reviewed:
- Description: This article presents a summary of the work done on the development of a new algorithm for mobile robot navigation in unknown dynamic environment. The developed humanlike algorithm uses a free-space attraction (Agoraphilic) concept for robot navigation. The algorithm presented in this article is an advanced development of the Agoraphilic navigation algorithm. The Agoraphilic algorithm does not look for obstacles (problems) to avoid but rather for free spaces toward the goal (solutions) to follow. The original Agoraphilic while it was able to overcome the limitations of the traditional algorithms had its own limitations in navigating robots in environments cluttered with moving obstacles. The new Agoraphilic Navigation Algorithm under Dynamic Environment (ANADE) was developed to overcome those limitations. ANADE consists of seven main modules reported in this article. The objects tracking and objects prediction methodologies developed for the algorithm estimate future growing free-space passages toward the goal. The algorithm generates a time-varying single attractive force to pull the robot through the free space toward the predicted (future) growing free-space passages leading to the goal. The new algorithm was tested, not only through simulation, but also through experimental work. Summary of the experimental results is presented and discussed in this article. © 1996-2012 IEEE.
Application of artificial intelligence in geotechnical engineering : a state-of-the-art review
- Authors: Baghbani, Abolfazl , Choudhury, Tanveer , Costa, Susanga , Reiner, Johannes
- Date: 2022
- Type: Text , Journal article , Review
- Relation: Earth-Science Reviews Vol. 228, no. (2022), p.
- Full Text: false
- Reviewed:
- Description: Geotechnical engineering deals with soils and rocks and their use in engineering constructions. By their nature, soils and rocks exhibit complex behaviours and a high level of uncertainty in material modelling. Artificial intelligence (AI) methods have been developed and used by an increasing number of researchers in the field of geotechnical engineering in the last three decades. These methods have been considered successful due to their ability to predict complex nonlinear relationships. Based on more than one thousand (i.e. 1235) published literatures, this paper presents a detailed review of the performance of AI methods and algorithms used in geotechnical engineering. Nine key areas where the application of AI methods is prominent were identified: frozen soils and soil thermal properties, rock mechanics, subgrade soil and pavements, landslide and soil liquefaction, slope stability, shallow and piles foundations, tunnelling and tunnel boring machine, dams, and unsaturated soils. Artificial Neural Network (ANN) emerged as the most widely used and preferred AI method with 52% of studies relying on it. Other methods that were used to a lesser extent were FIS, ANFIS, SVM, LSTM, CNN, ResNet and GAN. The analysis shows that the success and accuracy of AI applications depends on the number and type of datasets and selection of input parameters. The paper also provides statistical information on research incorporating AI methods and discusses the opportunities and challenges for future research and practical applications in geotechnical engineering. © 2022 Elsevier B.V.
Smart textiles for improved quality of life and cognitive assessment
- Authors: Oatley, Giles , Choudhury, Tanveer , Buckman, Paul
- Date: 2021
- Type: Text , Journal article
- Relation: Sensors Vol. 21, no. 23 (2021), p.
- Full Text:
- Reviewed:
- Description: Smart textiles can be used as innovative solutions to amuse, meaningfully engage, comfort, entertain, stimulate, and to overall improve the quality of life for people living in care homes with dementia or its precursor mild cognitive impairment (MCI). This concept paper presents a smart textile prototype to both entertain and monitor/assess the behavior of the relevant clients. The prototype includes physical computing components for music playing and simple interaction, but additionally games and data logging systems, to determine baselines of activity and interaction. Using microelectronics, light-emitting diodes (LEDs) and capacitive touch sensors woven into a fabric, the study demonstrates the kinds of augmentations possible over the normal manipulation of the traditional non-smart activity apron by incorporating light and sound effects as feedback when patients interact with different regions of the textile. A data logging system will record the patient’s behavioral patterns. This would include the location, frequency, and time of the patient’s activities within the different textile areas. The textile will be placed across the laps of the resident, which they then play with, permitting the development of a behavioral profile through the gamification of cognitive tests. This concept paper outlines the development of a prototype sensor system and highlights the challenges related to its use in a care home setting. The research implements a wide range of functionality through a novel architecture involving loosely coupling and concentrating artifacts on the top layer and technology on the bottom layer. Components in a loosely coupled system can be replaced with alternative implementations that provide the same services, and so this gives the solution the best flexibility. The literature shows that existing architectures that are strongly coupled result in difficulties modeling different individuals without incurring significant costs. © 2021 by the authors. Licensee MDPI, Basel, Switzerland.
Non-functional regression : A new challenge for neural networks
- Authors: Vamplew, Peter , Dazeley, Richard , Foale, Cameron , Choudhury, Tanveer
- Date: 2018
- Type: Text , Journal article
- Relation: Neurocomputing Vol. 314, no. (2018), p. 326-335
- Full Text:
- Reviewed:
- Description: This work identifies an important, previously unaddressed issue for regression based on neural networks – learning to accurately approximate problems where the output is not a function of the input (i.e. where the number of outputs required varies across input space). Such non-functional regression problems arise in a number of applications, and can not be adequately handled by existing neural network algorithms. To demonstrate the benefits possible from directly addressing non-functional regression, this paper proposes the first neural algorithm to do so – an extension of the Resource Allocating Network (RAN) which adds additional output neurons to the network structure during training. This new algorithm, called the Resource Allocating Network with Varying Output Cardinality (RANVOC), is demonstrated to be capable of learning to perform non-functional regression, on both artificially constructed data and also on the real-world task of specifying parameter settings for a plasma-spray process. Importantly RANVOC is shown to outperform not just the original RAN algorithm, but also the best possible error rates achievable by any functional form of regression.
Modular implementation of artificial neural network in predicting in-flight particle characteristics of an atmospheric plasma spray process
- Authors: Choudhury, Tanveer , Berndt, Christopher , Man, Zhihong
- Date: 2015
- Type: Text , Journal article
- Relation: Engineering Applications of Artificial Intelligence Vol. 45, no. (2015), p. 57-70
- Full Text: false
- Reviewed:
- Description: This paper presents a modular implementation of an artificial neural network to model the atmospheric plasma spray process in predicting the in-flight particle characteristics from the input processing parameters. The in-flight particle characteristics influence the structure and properties of the thermal spray coating and, thus, are considered important parameters to comprehend, simulate and predict the manufacturing process. The modular implementation allows simplification of the optimized model structure with enhanced ability to generalise the network. As well, the underlying relationship between each of the output in-flight characteristics with respect to the input processing parameters is explored. Smaller networks are constructed that achieves better, or in some cases, similar results. The training process is found to be more robust and stable along with fewer fluctuations in the values of the network parameters. The networks also respond to the variations of the number of hidden layer neurons with some definite trend. The predictable trend enhances reliability of the application of the artificial neural network in modelling the atmospheric plasma spray process and overcomes the variability and non-linearity associated with the process. © 2015 Elsevier Ltd. All rights reserved.
An extreme learning machine algorithm to predict the in-flight particle characteristics of an atmospheric plasma spray process
- Authors: Choudhury, Tanveer , Berndt, Christopher , Man, Zhihong
- Date: 2013
- Type: Text , Journal article
- Relation: Plasma Chemistry and Plasma Processing Vol. 33, no. 5 (2013), p. 993-1023
- Full Text: false
- Reviewed:
- Description: A robust single hidden layer feed forward neural network (SLFN) is used in this study to model the in-flight particle characteristics of the atmospheric plasma spray (APS) process with regard to the input processing parameters. The in-flight particle characteristics influence the structure and properties of the APS coating and, thus, are considered important parameters to comprehend the manufacturing process. The training times of traditional back propagation algorithms, mostly used to model such processes, are far slower than desired for implementation of an on-line control system. Use of slow gradient based learning methods and iterative tuning of all network parameters during the learning process are the two major causes for such slower learning speed. An extreme learning machine (ELM) algorithm, which randomly selects the input weights and biases and analytically determines the output weights, is used in this work to train the SLFNs. Performance comparisons of the networks trained with ELM algorithm and standard error back propagation algorithms are presented. It is found that networks trained with ELM have good generalization performance, much shorter training times and stable performance with regard to the changes in number of hidden layer neurons. The trends represent robustness of the trained networks and enhance reliability of the application of the artificial neural network in modelling APS processes. © 2013 Springer Science+Business Media New York.
Applicability of artificial neural network in hydraulic experiments using a new sewer overflow screening device
- Authors: Aziz, Md Abdul , Imteaz, Monzur , Choudhury, Tanveer , Phillips, David
- Date: 2013
- Type: Text , Journal article
- Relation: Australian Journal of Water Resources Vol. 17, no. 1 (2013), p. 77-86
- Full Text: false
- Reviewed:
- Description: During wet weather conditions, sewer overflows to receiving water bodies raise serious environmental, aesthetic and public health problems. These issues trigger the need the most appropriate device/system for a particular installation, especially at unmanned remote locations. A new sewer overflow device consists of a rectangular tank and a sharp crested weir with a series of vertical combs is presented. A series of laboratory tests to determine trapping efficiencies for common sewer solids were conducted for different flow conditions, number of combs layers and spacing of combs. To overcome physical limitations inherent in laboratory studies such as significant cost and time. Artificial neural model was adopted as it has the capacity to accurately predict the outcome of complex, non-linear physical systems with relatively poorly understood physicochemical processes. A series of laboratory tests were conducted with 55 different sets of data. Forty-seven sets of experimental data are used with 60% for training, 20% each for testing and validation of the model. A separate validation data sets were used to judge the overall performance of the trained network. The model can successfully predict the experimental results with more than 90% accuracy with an average absolute percentage error of around 7%. © Institution of Engineers Australia, 2013.
Artificial Neural Network application for predicting in-flight particle characteristics of an atmospheric plasma spray process
- Authors: Choudhury, Tanveer , Hosseinzadeh, Nasser , Berndt, Christopher
- Date: 2011
- Type: Text , Journal article
- Relation: Surface and Coatings Technology Vol. 205, no. 21-22 (2011), p. 4886-4895
- Full Text: false
- Reviewed:
- Description: Thermal spray consists of a group of coating processes that are used to apply metal or non-metallic coatings to protect a functional surface or to improve its performance. There are some 40 processing parameters that define the overall coating quality and these must be selected in an optimized fashion to manufacture a coating that exhibits desirable properties. The proper combination of processing variables is critical since these influence the cost as well as the coating characteristics.Because of this high number of processing parameters, a major challenge is to have full control over the system and to understand parameter interdependencies, correlations and their individual effects on the in-flight particle characteristics, which have significant influence on the in service coating properties. This paper proposes an approach, based on the Artificial Neural Network (ANN) method, to play this role and illustrates the model's design, network optimization procedures, the database handling and expansion steps, and analysis of the predicted values, with respect to the experimental ones, in order to evaluate the network's performance. © 2011 Elsevier B.V.