A Recurrent Neural-Network-Based Real-Time Dynamic Model for Soft Continuum Manipulators

Numerous soft and continuum robotic manipulators have demonstrated their potential for compliant operation in highly unstructured environments or near people. Despite their recent popularity, modeling of their smooth bending deformation remains a challenge. For soft continuum manipulators, the widespread, constant curvature approach to modeling is inadequate for modeling some deformations that occur in practice, such as combined bending and twisting deformations. In this paper, we extend the classical Cosserat rod approach to model a variable-length, pneumatic soft continuum arm. We model the deformation of a pneumatically driven soft continuum manipulator, and the model is then compared against experimental data collected from a three degree of freedom, pneumatically actuated, soft continuum manipulator. The model shows good agreement in capturing the overall behavior of the bending deformation, with mean Euclidean error at the tip of the robot of 2.48 cm for a 22 cm long robot. In addition, the model shows good numerical stability for simulating long duration computations.

Position regulation in Cartesian space of a class of inextensible soft continuum manipulators with pneumatic actuation

Planar soft space robotic manipulators: Dynamic modeling and control

Soft Continuum manipulators are increasingly popular because of their ability to interact safely with humans, maneuver around obstacles and enable cost-effective operation. In this paper, we investigate a unique manipulator composed of two asymmetric fiber reinforced pneumatic building blocks, one capable of bending and the other axially rotating. The combination of these building blocks can yield spatial motion including helical deformation pattern. While this design architecture is beneficial for whole arm manipulation and grasping, the asymmetric combination introduces structural coupling between the two modes rendering an analytical model hard to formulate. In this paper, we propose a Cosserat rod model to capture the deformation of the manipulator whose material properties and actuation parameters are estimated through extensive experiments. Once the model parameters are determined, the Cosserat model can be used to estimate deformation profile even in the presence of external loads. Such a framework is general and can be applied to obtain the forward and inverse kinematics of any continuum manipulator.

Continuum manipulators are continuously bending robots with unlimited number of kinematic degrees of freedom. Most existing continuum manipulators have a central strut made of a single elastic material, and multiple cables placed around the strut are employed to actuate the manipulator. The kinematics for such robots has been extensively studied by assuming the manipulator has a circular shape. In this paper, we aim to investigate the kinematic and static modeling for novel soft continuum manipulators fabricated by multi-material 3D printing with heterogeneous soft materials. We model cylindrical shape manipulators consisting of three sections with three different materials, and they are actuated by three independent cables placed symmetrically. By pulling the cables with different displacements, the manipulators can be bent in three-dimensional space. As our initial study, we employ a single cable for all prototypes. Experimental results are compared with the simulation results to validate the proposed modeling method.

PurposeThe current methods for inspecting tall or deep structures such as towers, chimneys, silos, and wells suffer from certain constraints. Manual and assisted inspection methods including humans, drones, wall climbing robots, and others are either costly, have a limited operation time, or affected by field conditions, such as temperature and radiation. This study aims to overcome the presented challenges through a teleoperated soft continuum manipulator capable of inspecting tall or deep structures with high resolution, an unlimited operation time and the ability to use different arms of the manipulator for different environments and structure sizes.Design/methodology/approachThe teleoperated manipulator uses one rotary and two tendon actuators to reach and inspect the interior of a tall (or deep) structure. A sliding part along the manipulator’s body (arm constrainer and tendon router) induces a variable-length bending segment, allowing an inspection camera to be placed at different distances from the desired location.FindingsThe experiments confirmed the manipulator’s ability to inspect different locations in the structure’s interior. The manipulator also demonstrated a submillimeter motion resolution vertically and a 2.5 mm per step horizontally. The inspection time of the full structure was 48.53 min in the step-by-step mode and was calculated to be 4.23 min in the continuous mode.Originality/valueThe presented manipulator offers several design novelties: the arm’s thin-wide cross-section, the variable-length bending segment in a fixed-length body, the external rolling tendon routing and the ability to easily replace the arm with another of different material or dimensions to suite different structures and environments.

Soft robotics offer biocompatibility, dexterity, and safe tissue interaction in surgery, providing potential alternatives to conventional tools such as colonoscopes. However, their nonlinear behaviour demands closed-loop control with structure-compatible feedback. This work presents a scalable 3D shape-sensing method based on Electrical Impedance Tomography (EIT), unifying actuation and sensing of a hydraulically actuated soft robot within a neural network framework. A soft continuum manipulator (14.6 mm in diameter) with saline-pressurised chambers and embedded kirigami-inspired FPCs was evaluated in free motion and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">ex vivo</i> porcine colon trials. A multilayer perceptron (MLP) predicted the full 3D shape, achieving tip RMSEs of 0.46, 0.20, and 0.40 mm (x, y, z) in free motion, and 1.96, 0.86, and 0.89 mm in <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">ex vivo</i>. This paper marks the first <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">ex vivo</i> validation of EIT-based shape sensing in soft endoscopy and demonstrating its potential for closed-loop surgical control.

The Sixth International Symposium on Neural Networks (ISNN 2009)

Besides the feedforward neural networks, there are the recurrent networks, where the impulses can be transmitted in both directions due to some reaction connections in these networks. Recurrent Neural Networks (RNNs) are linear or nonlinear dynamic systems. The dynamic behavior presented by the recurrent neural networks can be described both in continuous time, by differential equations and at discrete times by the recurrence relations (difference equations). The distinction between recurrent (or dynamic) neural networks and static neural networks is due to recurrent connections both between the layers of neurons of these networks and within the same layer, too. The aim of this chapter is to describe a Recurrent Fuzzy Neural Network (RFNN) model, whose learning algorithm is based on the Improved Particle Swarm Optimization (IPSO) method.

This research aims at developing a real-time simulation-based dynamic traffic model for mixed traffic flow conditions for ATMS/ATIS applications. The real-time model is implemented based on rolling horizon approach. The rolling horizon is advanced for each stage, thus real-time data can be incorporated within the framework. In order to predict normative as well as predictive information, a simulation-based DTA model is employed within each stage. Empirical data for freeways, including flows from toll stations and vehicle detectors, are used to validate the model in a real-time environment.

Recurrent neural networks (RNNs) can be used to handle sequential patterns and have been used for speech recognition. To overcome the shortcomings of RNN, recurrent sub neural networks (RSNNs) are used, where an RSNN is built independently for each class. The training algorithm of the RSNN is based on the backpropagation algorithm. Speaker dependent connected Chinese digit-speech recognition experiments were carried out. Some factors influencing the performance of RSNNs have been studied. The experiments show that RSNN is easier to train and gives higher performance than RNN.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

This research aims at developing a system of real-time simulation-based dynamic traffic model under mixed traffic flow conditions. The system consists of two layers, namely, simulation layer and real-time control layer. The system is implemented based on rolling horizon approach, which is advanced for each stage; thus real-time data can be incorporated within the framework. In order to predict normative, as well as predictive, information, a simulation-based dynamic traffic assignment model is employed within each stage. Empirical data for a real city network, such as flows from vehicle detectors, are used to validate the model in a real-time environment. The values of mean absolute percentage error and root-mean-squared percentage error are within 15%, and the results show promising agreements between observed and simulated flows.

Reversible lanes play an important role in alleviating urban traffic congestion and improving the utilization of road resources. However, most of the reversible lanes currently implemented are timed and fixed, and there is little research on the safety of reversible lanes without isolation facilities. In view of the existing problems of reversible lanes, this paper proposed a real-time dynamic reversible lane safety control model. The influencing factors of reversible lane safety control are analyzed first, and then determine model parameters of the no-entry zone and the buffer zone in the conflict area. Moreover, model parameters of emptying, exit, and recovery zones in the operation process were determined as well. Additionally, this paper discussed the length of reversible lanes and the timing of signals. The surrogate safety assessment model (SSAM) is adopted to analyze vehicle trajectory data for traffic simulation under different traffic flows. The experiment results show that the proposed reversible lane safety control scheme can keep the vehicle’s traffic conflict rate below 5%, which can effectively ensure the safe operation of the reversible lane.

Prediction concept needs human-like thinking and ideas. ANN helps make a program for describing the solutions to the prediction problems. To gain accurate prediction, there is a need for deep thinking in the knowledge system discovered. Deep learning helps form such next-level type of ANN thinking. Deep learning models do not require humans for programming the problems. It can determine the prediction by itself by learning the dataset. This paper describes the prediction by considering sequential model for continuous accessing of data with the help of multi-layers. Deep learning is very helpful for predicting larger dataset. This paper proposes an algorithmic model known as a Comprehensive Deep Recurrent Artificial Neural Networks (CDRANN) by hybridizing LSTM (long short-term memory) with RNN (recurrent neural network) backtrack solver for better future stock prediction. This paper also explains the need for hybridizing evolutionary model CDRANN to create a novel change in future prediction.KeywordsCDRANN (Comprehensive Deep Recurrent Artificial Neural Networks)LSTM (long short-term memory)RNN (recurrent neural network)Stock marketPrediction

Recurrent Neural Networks (RNNs), which are a powerful scheme for modeling temporal and sequential data need to capture long-term dependencies on datasets and represent them in hidden layers with a powerful model to capture more information from inputs. For modeling long-term dependencies in a dataset, the gating mechanism concept can help RNNs remember and forget previous information. Representing the hidden layers of an RNN with more expressive operations (i.e., tensor products) helps it learn a more complex relationship between the current input and the previous hidden layer information. These ideas can generally improve RNN performances. In this paper, we proposed a novel RNN architecture that combine the concepts of gating mechanism and the tensor product into a single model. By combining these two concepts into a single RNN, our proposed models learn long-term dependencies by modeling with gating units and obtain more expressive and direct interaction between input and hidden layers using a tensor product on 3-dimensional array (tensor) weight parameters. We use Long Short Term Memory (LSTM) RNN and Gated Recurrent Unit (GRU) RNN and combine them with a tensor product inside their formulations. Our proposed RNNs, which are called a Long-Short Term Memory Recurrent Neural Tensor Network (LSTMRNTN) and Gated Recurrent Unit Recurrent Neural Tensor Network (GRURNTN), are made by combining the LSTM and GRU RNN models with the tensor product. We conducted experiments with our proposed models on word-level and character-level language modeling tasks and revealed that our proposed models significantly improved their performance compared to our baseline models.