Abstract:

Abstract:Wheel is a great invention of human beings. The wheeled mobile system has brought great convenience to humans in production and daily life. Therefore, the wheeled robot is a very important development direction. To overcome the problem of insufficient adaptability of ordinary wheels to complex terrain, many research works have been investigated including the adhesion mode between wheels and supporting surface, the wheels’ geometry, the steering mode and so on. In this paper, based on the mechanism of wheels changing sliding friction into rolling friction, this paper proposes three new wheel structures, including tri-mode transformable wheel, magnetic adhesion wheel and variable parameter omni-directional wheel, and their corresponding wheel mobility structure, which aims at the requirements of uneven terrain crossing, vertical wall climbing and ground omni-directional movement. Through the linkage mechanism, the tri-mode transformable wheel switches its geometry among round mode, claw mode and hook mode, and the bi-directional obstacle crossing of the robot is realized; Through the magnetic adhesion wheel and passive suspension with three degrees of freedom, the robot can move on the wall with all wheels attached to the surface; Through applying the spatial mechanism to adjust the roller installation angles, the direction of wheel friction can be controlled and the omni-directional driving of the robot can be realized; through prototype construction and experimental study, the feasibility of the proposed innovative designs are verified.

Abstract:In order to solve the difficulty of multi-wall transition in the cleaning process of ship cargo hold,a magnetic adsorption wall-climbing robot with adaptive ability to wall is designed in this paper, which includes magnetic adsorption mechanism, adaptive cleaning mechanism and walking mechanism. Firstly, a mechanical model of the robot in the process of wall transition is established to obtain the distribution characteristics of magnetic adsorption force and a magnetic adsorption mechanism is designed. Secondly, ANSYS Maxwell 3D software is used to optimize the distribution of magnetic adsorption force of the mechanism to meet the needs of wall transition. An adaptive cleaning mechanism is designed in the head of the robot, and the wall transition ability of the cleaning mechanism is verified by experiments. Finally, the robot prototype is tested according to the actual characteristics of cargo hold wall. Experiments show that the wall-climbing robot can complete the transition process between the bilge and the roof of the cargo hold, which verifies the robot’s ability to adapt to the wall and walk in the cabin.

Abstract:It is still a challenge to design a hand prosthesis with a consideration of multi- motions and light weight. In this paper, by analyzing 16 commonly-used motions of human hands, a trade-off plan between weight saving and the number of motions of a hand prosthesis is studied. We determine the functions to be implemented as a constant interlock mechanism of four fingers. An adaptive mechanism is applied for the thumb, and the symmetric series elastic actuator is used for the arching of metacarpal. With such a design, a prosthetic hand was designed just using three motors embedded in the palm, which has weight of 132.1 g and could perform 11 motions. The grasping stability and operability of the hand prosthesis were confirmed with intuitive myoelectric control based on a neural network algorithm in the subject experiments.

Abstract:Underwater bionic robots have distinct advantages such as the high efficiency, high mobility and low noise etc. In this paper, a deep reinforcement learning based method is studied to control the robotic eel. Firstly, based on the propulsion principle of active and passive bionic mechanism, a robotic eel with two active rigid bodies and two compliant bodies is designed. Secondly, the robotic eel is modeled and simulated. The data collecting and training tasks are carried out in the simulation environment using deep reinforcement learning algorithms. The neural network with better performance is selected as the control function for the robotic eel. Finally, feasibility of the design and effectiveness of the control function are verified by a prototype via real experiments.

Abstract:To study the human-robot collaborative assembly process of space erectable truss structure, an innovative configuration that suitable for radial fast assembly of truss modules was proposed in this work. Based on state matrix and adjacency matrix, mathematic models of assembly sequence, assembly mode and assembly process of truss structures was established firstly. Then, the human-robot capability constraints of truss assembly in space environment were analyzed. The hierarchical decomposition of assembly tasks based on dynamic operation element analysis was proposed. The process and assignment scheme of human-robot collaborative assembly tasks suitable for programming were given by using the principle of comparison and allocation. Finally, to verify the rationality and feasibility of the proposed assembly process, a ground demonstration test was carried out for the collaborative assembly of 5 m long erectable truss structure by human wearing simulated space suit and manipulator. The results showed that, the proposed scheme can provide an effective way to realize the on-orbit construction of large space facilities.

Abstract:Mice are widely used in various physiological, pathological and behavioral experiments. However, it is yet unclear whether or how much spontaneous behavior varies between mice with different genotypes, which may affect experiment design and outcome. In this study, three inbred mice C57BL/6J, C57BL/6N, BALB/c (6J, 6N, BC) were selected, and objectively evaluated for their spontaneous behavior with a hierarchical unsupervised learning framework for 3D animal behavioral characteristics. The results showed that genotype is the most significant factors influencing mice, with the whole list presented as distant genotype＞near genotype＞gender difference. Behavior atlas varies greatly between BC and 6J/6N groups due to obvious differences in body posture, and BC have significantly lower movement speed than 6J and 6N. Although 6J and 6N behavior atlas were relatively similar, 6N sniffed significantly than 6J, and the anxiety levels at 6N were relatively high. The results of this study can be used as a reference for experiment design, strain selection, and determination of N value in future research using mouse model.

Abstract:In many automation application scenarios, such as assembly and sorting processes, the use of industrial robots is an important part of improving production quality and productivity. During the working processes, safety is one of the primary prerequisites to be considered.[ In this paper, a safety warning system based on 3D vision for robots is proposed. First, the system uses 3D cameras to perform high-precision 3D reconstruction of the monitoring scene and fuse multiple point clouds; secondly, the human key points are extracted and the safety distance between the human and the robot arm is judged according to the key|points; finally,the calculation determines whether the human and the robot are at the set safety distance and controls the working state of the robot arm. Experimental results show that the proposed safety warning system is able to work within a large field of view.

Abstract:The complex system of human body can be observed by observing the change of gait rhythm.The dynamic characteristics of the time series of stride interval during walking can effectively reflect the state change of human system, which can be used for abnormal gait detection and related disease identification.Time series phase space reconstruction of human gait sensor information is an effective modeling method to characterize nonlinear dynamics of the system. Geometric modeling and statistical analysis of phase space are typical analysis methods for abnormal gait recognition, which are widely used in clinical research such as neurodegenerative disease detection. In this paper, a topological nonlinear dynamic modeling method is proposed to identify abnormal gait in neurodegenerative diseases from the perspective of spatial topological analysis. Firstly, a phase-space reconstruction method based on time-delay embedding is used to transform the wave time series of gait into an abstract phase-space state point cloud. Secondly, the continuous homology tool based on computational topology is used to extract the topology description information of the space where the state point cloud resides. Thirdly, the topological nonlinear dynamic characteristics of time series are constructed by using the continuous situation graph based on topological description. Finally, a machine learning recognition model of abnormal gait was constructed by integrating the topological nonlinear dynamic characteristics of the time series of stride interval, standing interval and swing interval of left and right feet in the gait cycle as the input of classifier. The results showed that the area under receive-operator-curve was 0.875 0 (0.914 6), 0.940 6 (0.962 3) and 0.958 3 (0.961 4) in the 5-minute continuous walking data (50-step sliding window data) of abnormal gait in patients with neurodegenerative diseases of muscular sclerosis, Huntington’s disease and Parkinson’s disease, respectively. Therefore, topological nonlinear dynamic modeling analysis is an effective gait detection method for neurodegenerative diseases, and provides a new idea for neurodegenerative disease detection and wearable data analysis based on gait analysis.

Abstract: