Shenyang Institute of Automation,Chinese Academy Of Sciences
Not a member yet
23582 research outputs found
Sort by
具有冗余供电的空间站舱外移动式机器人系统
No full text本实用新型涉及载人航天空间站设备，特别涉及一种具有冗余供电的空间站舱外移动式机器人系统。包括移动平台、弧形轨道系统、收放线机构、空间站暴露平台舱段及六轴机械臂，其中弧形轨道系统设置于空间站主舱段和空间站暴露平台舱段之间，移动平台设置于弧形轨道系统上，六轴机械臂设置于移动平台上，移动平台可沿弧形轨道系统运动，从而实现六轴机械臂在空间站暴露平台舱段所有象限内的连续可达；收放线机构设置于弧形轨道系统的底部，且与移动平台连接，用于为移动平台供电。本实用新型采用摩擦轮驱动的移动平台作为机械臂的基座，使机械臂在暴露平台全象限可连续可达、定位，操作过程中不需要重新更换基准点，易于保证精度
Monocular 3D Object Detection Based on Uncertainty Prediction of Keypoints
No full textThree-dimensional (3D) object detection is an important task in the field of machine vision, in which the detection of 3D objects using monocular vision is even more challenging. We observe that most of the existing monocular methods focus on the design of the feature extraction framework or embedded geometric constraints, but ignore the possible errors in the intermediate process of the detection pipeline. These errors may be further amplified in the subsequent processes. After exploring the existing detection framework of keypoints, we find that the accuracy of keypoints prediction will seriously affect the solution of 3D object position. Therefore, we propose a novel keypoints uncertainty prediction network (KUP-Net) for monocular 3D object detection. In this work, we design an uncertainty prediction module to characterize the uncertainty that exists in keypoint prediction. Then, the uncertainty is used for joint optimization with object position. In addition, we adopt position-encoding to assist the uncertainty prediction, and use a timing coefficient to optimize the learning process. The experiments on our detector are conducted on the KITTI benchmark. For the two levels of easy and moderate, we achieve accuracy of 17.26 and 11.78 in AP(3D), and achieve accuracy of 23.59 and 16.63 in AP(BEV), which are higher than the latest method KM3D
Learning Cognitive Map Representations for Navigation by Sensory-Motor Integration
No full textHow to transform a mixed flow of sensory and motor information into memory state of self-location and to build map representations of the environment are central questions in the navigation research. Studies in neuroscience have shown that place cells in the hippocampus of the rodent brains form dynamic cognitive representations of locations in the environment. We propose a neural-network model called sensory-motor integration network model (SeMINet) to learn cognitive map representations by integrating sensory and motor information while an agent is exploring a virtual environment. This biologically inspired model consists of a deep neural network representing visual features of the environment, a recurrent network of place units encoding spatial information by sensorimotor integration, and a secondary network to decode the locations of the agent from spatial representations. The recurrent connections between the place units sustain an activity bump in the network without the need of sensory inputs, and the asymmetry in the connections propagates the activity bump in the network, forming a dynamic memory state which matches the motion of the agent. A competitive learning process establishes the association between the sensory representations and the memory state of the place units, and is able to correct the cumulative path-integration errors. The simulation results demonstrate that the network forms neural codes that convey location information of the agent independent of its head direction. The decoding network reliably predicts the location even when the movement is subject to noise. The proposed SeMINet thus provides a brain-inspired neural-network model for cognitive map updated by both self-motion cues and visual cues
Deep Object Detector With Attentional Spatiotemporal LSTM for Space Human–Robot Interaction
No full textGlobal temporal information and local semantic information are essential cues for high-performance online object detection in videos. However, despite their promising detection accuracy in most cases, most state-of-the-art approaches have following two limitations: invalid background/scale suppression and inadequate temporal information mining between frames. Many jobs currently focus on temporal information learning based on a single frame. In this article, we propose an attentional global–local information learning network; this is one of the first attempts to fully use both types of information between frames. Attention maps are creatively utilized to transfer temporal contexts between frames. This also effectively alleviates the adverse effects of scale changes. Furthermore, empowered by a detailed framework, a proposed detector effectively uses multilevel feature extraction. Given these contributions, the proposed detector achieves state-of-the-art performance on challenging benchmarks. Finally, practical experiments are conducted on a space human–robot interaction platform.</p
A Potential Game Approach for Decentralized Resource Coordination in Coexisting IWNs
No full textTo meet the requirements of various emerging manufacturing applications, multiple Industrial Wireless Networks (IWNs) are employed to operate in the same region. However, the limited communication resources inevitably incur interference in the time and frequency domains, which is known as the coexistence problem. Existing centralized mechanisms suffer from a low computational efficiency in a large-scale network scenario, and the globally shared information cannot be fully obtained in practice. To this end, we first design an incomplete information sharing protocol to clarify the decentralized coordination among coexisting IWNs. We then formulate the coexistence problem as a non-cooperative game, which is proven to be a potential game. In addition, considering the deterministic deadlines of data transmissions in industrial applications, we propose a Deadline-aware Incomplete-Information-based Decentralized Resource Coordination (DIIDRC) algorithm. We also mathematically analyze that the DIIDRC algorithm converges to a collision-free optimal schedule in a resource-sufficient scenario or a nearly-optimal schedule in a resource-insufficient scenario. We conduct extensive simulations to verify the effectiveness of the DIIDRC algorithm. Evaluation results show that the DIIDRC algorithm has obvious superiorities over existing works in terms of the convergence rate and schedulable ratio.</p
Foldable Units and Wing Expansion of the Oakleaf Butterfly During Eclosion
No full textEclosion is a rapid process of morphological changes in insects, especially for the wings of butterflies. The orange oakleaf butterfly (Kallima inachus) transits from pupae to adults with a 9.3 fold instant increase in the surface area of their wings. To explore the mechanism for the rapid morphological changes in butterfly wings, we analyzed changes in microstructures in the wings of K. inachus. We found that there were lots of micron-sized foldable units in the wings at the pupal stage. The foldable units could provide as much as 31.35 times of increase in wing surface area. During eclosion, foldable units were flattened sequentially and resulted in a rapid increase in wing surface areas. The unfolding process was regulated by the structures and layouts of wing veins. Based on our observation, foldable units play important roles in both deformation and stretching of wings. The foldable units of microstructures may provide mimics for simulating entities of large-deformational bionic structures with practical application.</p
Research on optimization method of routing buffer linkage based on Q-learning
No full textAbstract
According to the characteristics of the painting process of passenger car manufacturing enterprises, by formulating the routing buffer linkage rules based on the total renewal cost, the linkage process of the bus in the routing buffer is controlled, and an improved Q-learning (Q- The routing buffer of learning) algorithm quickly finds the optimal path method. According to the actual production situation, this method improves the dynamic parameters of the algorithm on the basis of the traditional Q-learning algorithm, and improves the optimization speed and accuracy of the algorithm by establishing the correlation between the work-in-process and its neighboring work-in-process in the current state. Through multiple sets of example simulation tests, the effectiveness of the Q-learning algorithm in solving the optimization problem of routing buffer linkage is verified
Research on High-precision Unmanned Underwater Vehicles Team Formation without Communication Based on Visual Positioning Technology
No full text水下机器人集群技术是目前水下机器人技术领域的发展热点之一。针对以往基于通信的水下机器人编队存在的编队精度低、队形保持困难等问题，提出了基于水下矢量光图案及视觉定位的水下集群编队方法，并通过水池试验分别验证了视觉定位以及水下密集编队的功能和相关指标。试验结果表明：水下视觉定位能够达到不大于3%的定位精度和不小于2Hz的定位频率，能够为水下机器人自主航行提供准确连续的控制输入。基于视觉定位的水下集群能够实现相互距离10 m以内的密集编队，编队精度不大于10%，较以往基于通信的编队方法有较大提升。</p
TA15 and inconel 625 bimetallic structures additive manufacturing and phased array ultrasonic testing
No full textAdditive manufacturing of bimetallic structures has become a hotspot in recent years. However, the interfaces of bimetallic structures are always the weaknesses, and cracks always occur at the interfaces. Three bimetallic samples of TA15 and Inconel 625 were manufactured by laser metal deposition, and the effect of laser power on the manufacturing quality was analyzed. Then phased array ultrasonic testing (PAUT) technique was applied for crack detection for the three samples, and metallographic analysis was employed to verify detection results. The results showed that the increase of laser power can eliminate macro-cracks at the interface of the bimetallic sample, and micro-cracks at the interface can be detected by PAUT.</p
基于增材制造后处理的谢尔宾斯基曲线路径规划方法
No full text本发明涉及一种适用于增材制造加工后，进行减材精加工的基于谢尔宾斯基曲线的路径规划方法，包括：根据模型的几何特征进行映射区域分类；映射平面内求取谢尔宾斯基路径；将谢尔宾斯基路径映射到曲面上，获得精加工路径。本发明方法对复杂增材制造模型采用先进的减材制造路径，提高加工的无序性、消除中频、高频误差并消除抛光纹理；提高了加工效率和质量；采用根据曲面的几何特征进行映射区域类型的判断，选用合适的映射区域，保证不存在未加工到的残留区域以及重复区域；并根据去除精度设定阶数，保证质量的前提下要求加工路径短、效率高；该路径规划方式有效防止单一方向运行，消除中频、高频误差，实现高效的精密加工