Enguang Guan

A novel design for the self-reconfigurable robot module and connection mechanism

Many researchers have paid more and more attention to the lattice self-reconfigurable modular robot for its excellent flexibility in connection and separation movements of modules. The paper proposes a novel design of self-reconfigurable robot module (M-Lattice) with the overall description. Moreover, the key features of modules genderless pin-slot-based connection mechanism are introduced. Finally, the connection mechanism is proved to be feasible for the tasks of modules self-reconfiguration by the experiments.

M-Lattice: A self-configurable modular robotic system for composing space solar panels:

This article presents a modular robotic system called M-Lattice usable for automatically composing huge solar cell panels for space solar power satellites. The module structure design along with the model structure, the connector mechanism and the pattern of multi-module movement are described. A distributed control strategy called simulated growth enables the self-configuration. Using local communications, each module can execute path searching, path planning and multi-module movement. Computational simulations have demonstrated the practicability of the design. Whereas the current designs are mostly restricted to the situations when the modules are taken into the workspace one by one, the simulated growth strategy enables movement of several modules simultaneously or in a pipeline order.

Fault Self-Diagnosis for Modular Robotic Systems Using M-Lattice Modules

In the domain of modular robotic systems, self-configuration, self-diagnosis and self-repair are known to be highly challenging tasks. This paper presents a novel fault self-diagnosis strategy which consists of two parts: fault detection and fault message transmission. In fault detection, a bionic synchronization ‘healthy heartbeat’ method is used to guarantee the high efficiency of the exogenous detection strategy. For fault message transmission, the Dijkstra method is modified to be capable of guiding the passage of fault messages along the optimal path. In a modular robotic system, fault message transmission depends mainly on local communications between adjacent modules, so there is no need for global broadcast information. Computational simulations of one system form, M-Lattice, have demonstrated the practical effectiveness of the proposed strategy. The strategy should be applicable in modular robotic systems in general.

Self-reconfiguration path planning design for m-lattice robot based on genetic algorithm

M-Lattice is a kind of lattice modular robot, which can finish self-reconfiguration in three-dimensional plane. How to substitute the broken modules effectively is a critical question for modular robot system. In order to solve it, we introduce the topology structure of M-Lattice system and math representation for the reconfiguration question. An energy factor to illustrate the relationship between energy cost and moving path is defined. The non-real time path planning based on genetic algorithm is also given. From the results of simulation, the reliability and feasibility of the planning is demonstrated.

Meta-module Self-configuration Strategy for Modular Robotic System

Modular robotic system (MRS) consisting of several identical modules is able to adapt to situations by configuring to the morphology which best suits the environment. Utilizing this characteristic, this paper presents a self-configuration strategy for the M-Lattice MRS. Based on this strategy, the construction task under complex environment is converted to a self-configuration task accomplished by modular robots. Meta-module method is utilized to avoid system cracking. Meanwhile, a gradient greedy method in a virtual gravity field is selected to indicate the locomotion of meta-modules. Computational simulations demonstrate the feasibility and scalability of the proposed strategy.

Melanoma Segmentation and Classification in Clinical Images Using Deep Learning

In this paper, a deep learning computer aided diagnosis system (CADs) is proposed for automatic segmentation and classification of melanoma lesions, containing a fully convolutional neural network (FCN) and a specific convolutional neural network (CNN). FCN, which consists of a 28-layer neural structure, is designed for segmentation and with a mask for region of interest (ROI) as its output. Later, the CNN only uses the segmented ROI of raw image to extract features, while the DLCM features, statistical and contrast location features extracted from same ROI are merged into CNN features. Finally, the combined features are utilized by the fully connected layers in CNN to obtain the final classification of melanoma, malignant or benign. The training of FCN and CNN are separated with different loss functions. Publicly available database ISBI 2016 is used for evaluating the effectiveness, efficiency, and generalization capability with evaluating indicator, such as accuracy, precision, and recall. Preprocessing methods, such as data argumentation and balancing are utilized to make further improvements to performance. Experiments on a batch size of 100 images yielded an accuracy of 92%, a specificity of 93% and a sensitivity of 94%, revealing that the proposed system is superior in terms of diagnostic accuracy in comparison with the state-of-the-art methods.

Benign and malignant mammographic image classification based on Convolutional Neural Networks

Computerized breast cancer diagnosis system has played an import role in early cancer diagnosis. For this purpose, we apply deep learning by using convolutional neural networks (CNN) to classify abnormalities, benign or malignant, in mammographic images based on the mini Mammographic Image Analysis Society (mini-MIAS) database. Accuracy, sensitivity, and specific values are observed to evaluate the performance of the CNN. To improve the performance, we utilize image-preprocessing methods containing cropping, global contrast normalization, augmentation, local histogram equalization, and balancing preprocessing. We built four CNN models to study the impact of depth and hidden layer structure on model performance. The CNN-4d model performs best among four proposed CNN models consisting of four convolution layers with a dropout of 0.7. The CNN-4d model achieved a balance of high sensitivity (90.63%) and high specificity (87.67%), and an accuracy of 89.05%. The result of this study indicates that CNNs have promising potential in the field of intelligent medical image diagnosis.

Linearity of the Force Leverage Mechanism Based on Flexure Hinges.

This paper proposes development of a force leverage mechanism based on the flexure hinges. The primary function of this leverage mechanism is to transform an objective unbalance force/moment to a force sensor in the static unbalance measure system. The measure precision is dependent on the linearity of the force transmission of the force leverage mechanism. The kinematics of the force leverage mechanism is modeled based on the elastic model. The finite element method is used to verify the analytical solutions. Moreover, the effect of the initial external load on the linearity is investigated. Further, the virtual experiment is carried on to verify the linearity and sensitivity. The static unbalance measure system employing the proposed leverage mechanism has the advanced sensitivity of less than 0.03 gcm and performs excellent linearity.

Optimal Motion Planning for Mobile Welding Robot

This paper focuses on the motion planning method for a novel mobile welding robot (MWR), based on the screw theory. The robot consists of a vehicle unit and a 5-DOF manipulator, which equipped a torch at the end of manipulator. In order to finish the welding task, the kinematic motion planning strategy is of great importance. As the traditional strategy which uses inverse kinematic and polynomial interpolation may cause a waste of computing time, the screw theory is chosen to improve the strategy. From the simulation and experiment results, it can be found that the optimal motion planning method is reliable and efficient.

Piezoelectric Micro-Pump Suction Cup Design and Research on the Optimal Static Driving Characteristics

This paper first introduces the design and principles of a negative pressure suction cup with integrated piezoelectric valveless micro-pump. Laser engraving technique is used to build the multilayer prototype. In order to improve the adsorption performance of the suction cup, the static driving characteristics of the piezoelectric actuator have been optimized: based on a series of assumptions according to piezoelectric composite oscillators, a theoretical mechanics model of circular single-chip piezoelectric actuator has been constructed to explore the lateral deformation of the piezoelectric actuator in the effect of simple electric load. By the method of finite element simulation, the model is verified. Further, through optimizing its geometry design, the piezoelectric actuator has been enabled of the best static deformation characteristics.