Adaptive Computational Techniques Research Articles

Design and Modeling of an Intelligent Robotic Gripper Using a Cam Mechanism with Position and Force Control Using an Adaptive Neuro-Fuzzy Computing Technique

Manufacturers increasingly turn to robotic gripper designs to improve the efficiency of gripping and moving objects and provide greater flexibility to these objects. Neuro-fuzzy techniques are the most widespread in developing gripper designs. In this study, the traditional gripper design is modified by adding a suitable cam that makes it compatible with the basic design, and an adaptive neuro-fuzzy inference system (ANFIS) is used in a MATLAB Simulink environment. The developed gripper investigates the follower path concerning the cam surface curve, and the gripper position is controlled using the developed ANFIS-PID. Three methods are examined in the developed ANFIS-PID controller: grid partitioning (genfis1), subtractive clustering (genfis2), and fuzzy C-means clustering (genfis3). The results show that the added cam can improve the gripping strength and that the ANFIS-PID model effectively handles the rise time and supported settling time. The developed ANFIS-PID controller demonstrates more efficient performance than Fuzzy-PID and traditional tuned-PID controllers. This proposed controller does not achieve any overshoot, and the rise time is improved by approximately 50–51%, and the steady-state error is improved by 75–95%, compared with Fuzzy-PID and tuned PID controllers. Moreover, the developed ANFIS-PID controller provides more stability for a wide range of set point displacements—0.05 cm, 0.5 cm, and 1.5 cm—during the testing period. The developed ANFIS-PID controller is not affected by disturbance, making it well suited for robotic gripper designs. Grip force control is also investigated using the proposed ANFIS-PID controller and compared with the Fuzzy-PID in three scenarios. The result from this force control proves objects’ higher actual gripping performance by using the proposed ANFIS-PID.

Adaptive 3D descattering with a dynamic synthesis network

Deep learning has been broadly applied to imaging in scattering applications. A common framework is to train a descattering network for image recovery by removing scattering artifacts. To achieve the best results on a broad spectrum of scattering conditions, individual “expert” networks need to be trained for each condition. However, the expert’s performance sharply degrades when the testing condition differs from the training. An alternative brute-force approach is to train a “generalist” network using data from diverse scattering conditions. It generally requires a larger network to encapsulate the diversity in the data and a sufficiently large training set to avoid overfitting. Here, we propose an adaptive learning framework, termed dynamic synthesis network (DSN), which dynamically adjusts the model weights and adapts to different scattering conditions. The adaptability is achieved by a novel “mixture of experts” architecture that enables dynamically synthesizing a network by blending multiple experts using a gating network. We demonstrate the DSN in holographic 3D particle imaging for a variety of scattering conditions. We show in simulation that our DSN provides generalization across a continuum of scattering conditions. In addition, we show that by training the DSN entirely on simulated data, the network can generalize to experiments and achieve robust 3D descattering. We expect the same concept can find many other applications, such as denoising and imaging in scattering media. Broadly, our dynamic synthesis framework opens up a new paradigm for designing highly adaptive deep learning and computational imaging techniques.

Computation and analysis for airflow characteristics of physical quantities caused by nasal obstruction

Nasal obstruction is a common disease found in humans and one of the most popular research topics. This article proposes an adaptive computing algorithm to construct the mesh structure for simulating the unsteady turbulent airflow in the nasal cavity of patients with nasal airway obstruction (NAO) with the help of MIMICS 16.0 × 64 and COMSOL 5.4. The computing results provide a lucid explanation for the aerodynamic impairment of the nasal passages and efficiently locate the three-dimensional region of the obstruction. Details of the time-dependent flow properties in different nasal cavities indicate that the pressure gradient along the streamline provides the NAO region. Several physical properties were discussed during one respiratory cycle. The merits of the adaptive computing technique showed that the time consumption and precision of the solutions were both satisfactory. The implementation of computer-aided diagnosis can be expected with more efforts in the near future.

A survey of adaptive computational techniques for approximating dissipating differential equations

A survey of adaptive computational techniques for approximating dissipating differential equations

A Novel Statistical Model to Estimate Host Genetic Effects Affecting Disease Transmission.

There is increasing recognition that genetic diversity can affect the spread of diseases, potentially affecting plant and livestock disease control as well as the emergence of human disease outbreaks. Nevertheless, even though computational tools can guide the control of infectious diseases, few epidemiological models can simultaneously accommodate the inherent individual heterogeneity in multiple infectious disease traits influencing disease transmission, such as the frequently modeled propensity to become infected and infectivity, which describes the host ability to transmit the infection to susceptible individuals. Furthermore, current quantitative genetic models fail to fully capture the heritable variation in host infectivity, mainly because they cannot accommodate the nonlinear infection dynamics underlying epidemiological data. We present in this article a novel statistical model and an inference method to estimate genetic parameters associated with both host susceptibility and infectivity. Our methodology combines quantitative genetic models of social interactions with stochastic processes to model the random, nonlinear, and dynamic nature of infections and uses adaptive Bayesian computational techniques to estimate the model parameters. Results using simulated epidemic data show that our model can accurately estimate heritabilities and genetic risks not only of susceptibility but also of infectivity, therefore exploring a trait whose heritable variation is currently ignored in disease genetics and can greatly influence the spread of infectious diseases. Our proposed methodology offers potential impacts in areas such as livestock disease control through selective breeding and also in predicting and controlling the emergence of disease outbreaks in human populations.

Adapting to life: ocean biogeochemical modelling and adaptive remeshing

Abstract. An outstanding problem in biogeochemical modelling of the ocean is that many of the key processes occur intermittently at small scales, such as the sub-mesoscale, that are not well represented in global ocean models. This is partly due to their failure to resolve sub-mesoscale phenomena, which play a significant role in vertical nutrient supply. Simply increasing the resolution of the models may be an inefficient computational solution to this problem. An approach based on recent advances in adaptive mesh computational techniques may offer an alternative. Here the first steps in such an approach are described, using the example of a simple vertical column (quasi-1-D) ocean biogeochemical model. We present a novel method of simulating ocean biogeochemical behaviour on a vertically adaptive computational mesh, where the mesh changes in response to the biogeochemical and physical state of the system throughout the simulation. We show that the model reproduces the general physical and biological behaviour at three ocean stations (India, Papa and Bermuda) as compared to a high-resolution fixed mesh simulation and to observations. The use of an adaptive mesh does not increase the computational error, but reduces the number of mesh elements by a factor of 2–3. Unlike previous work the adaptivity metric used is flexible and we show that capturing the physical behaviour of the model is paramount to achieving a reasonable solution. Adding biological quantities to the adaptivity metric further refines the solution. We then show the potential of this method in two case studies where we change the adaptivity metric used to determine the varying mesh sizes in order to capture the dynamics of chlorophyll at Bermuda and sinking detritus at Papa. We therefore demonstrate that adaptive meshes may provide a suitable numerical technique for simulating seasonal or transient biogeochemical behaviour at high vertical resolution whilst minimising the number of elements in the mesh. More work is required to move this to fully 3-D simulations.

Aggregating descriptive regularization and Bayesian nonparametric spectral estimation approaches for enhanced radar imaging

In this paper, we address and discuss a novel look at the high-resolution array radar/SAR imaging as an ill-conditioned inverse spatial spectrum pattern (SSP) estimation problem. The system-oriented theoretical developments are addressed to as an aggregated descriptive regularization-Bayesian (DRB) method for radar/SAR image formation/reconstruction. We exemplify how this aggregated method leads to new robust adaptive computational techniques that enable one to derive efficient and consistent estimates of the SSP via unifying the Bayesian minimum risk nonparametric spectral estimation strategy with the maximum entropy randomized a priori image model and other projection-type regularization constraints imposed on the solution. The reported simulation results demonstrate the efficiency of the addressed DRB-related radar/SAR-oriented enhanced imaging techniques.

A two-dimensional thin-film transistor simulation using adaptive computing technique

Abstract In this paper, an adaptive computational technique is applied to solve a set of two-dimensional (2D) drift-diffusion (DD) equations together with nonlinear trap model in thin-film transistors (TFTs). Different from the conventional DD equations in metal-oxide-semiconductor field effect transistors, the nonlinear trap model depending on the potential energy accounts for the effect of grain boundary on the electrical characteristics of low temperature polycrystalline-silicon (LTPS) TFTs. Our adaptive computing technique is mainly based on Gummel’s decoupling method, a finite volume (FV) approximation, a monotone iterative (MI) method, a posteriori error estimation, and an 1-irregular meshing scheme. Applying Gummel’s decoupling method to the set of DD equations firstly, each decoupled partial differential equation (PDE) is then approximated with FV method over 1-irregular mesh. Instead of conventional Newton’s iterative method, the corresponding system of nonlinear algebraic equations is solved with MI method. Variations of the computed solutions, such as potential and electron density are captured and a posteriori error estimation scheme is adopted to assess the quality of the computed solutions. The mesh is adaptively refined accordingly. The numerical method converges monotonically in both MI and Gummel’s iteration loops, respectively. Various cases of simulation have been verified for a typical LTPS TFT to demonstrate the accuracy and robustness of the method.

Knowledge based and adaptive computational techniques for concurrent design of powder metallurgy parts

Abstract The practice of concurrent engineering (CE) is adopted widely to facilitate integrated design and manufacture by industry in order to maintain competitiveness in the market place. Use of near net-shape processes like powder metallurgy (PM) within a CE environment can give added benefits in terms of material utilisation and environmental considerations. A system for concurrent design of PM parts is described here. It is developed to assist material and process selection while optimising the design for a part to be manufactured using the cold-compaction process. The system uses a rule based design geometry evaluations program to assess the parts' suitability for cold compaction. Design data in B-rep format conforming to the STEP standard is employed in this work. Material and process parameter selections are performed using a hybrid system employing Bayesian neural networks and heuristics. By employing an integrated system to aid material selection and manufacturability analysis during the early design phase, part designers can reduce the number of design iterations, improve quality of designs, and minimise trials required in PM part manufacture.