基于树莓派的手势交互六足机器人

将树莓派和YOLO v3目标检测算法相结合,提出了一种新型的六足机器人人机交互模式,使六足机器人完成五种不同动作(前进、后退、左拐、右拐、步态切换).其中室内与室外的手势识别在测试集上的准确率分别可达95.08％与89％.相比基于人工提取特征的传统算法和其他基于卷积神经网络(CNN)的算法具有更高的识别准确率和更强的鲁...     

基于YOLOv5s的扫地机器人

扫地机器人是一款服务型自动化机器人,它的诞生使我们得以从繁琐的家居清扫劳作中解放出来,让我们可以在更关键的应用领域当中更有精神和活力,极大程度地提升了我们的生活质量。文章采用YOLOv5s目标检测算法对U盘、钥匙等生活中容易掉落的小物品进行识别,首先分析了YOLOv5s的架构和原理,然后通过YOLOv5s对标注好的数据集进行训练,最后得到准确的检测结果,反馈到扫地机器人单片机上,对小物品进行处理。实验结果体现出了YOLOv5s的检测快速性和准确性。     

基于树莓派和YOLOv5的PCB瑕疵检测

针对PCB瑕疵检测问题,提出了基于YOLOv5s的轻量化PCB瑕疵检测算法,并基于树莓派平台搭建了一套PCB瑕疵自动检测系统。首先,在Backbone阶段使用改进的空间金字塔池化代替原有的C3网络;其次,在Backbone与Neck中引入残差结构,并在小目标检测层面加入CBAM注意力机制;最后,将所提轻量化算法部署到树莓派上,并使用NCS2套件进行辅助加速,通过摄像头进行自动检测。通过测试,所提算法检测PCB瑕疵mAP达到99.1%,与原YOLOv5s模型相比,Params为其23%,FLOPs为其21%,PCB瑕疵检测系统运行速度达到7 fps,满足自动检测要求。     

基于改进YOLOv5的服务机器人目标识别算法

在服务机器人执行任务的过程中,对周围环境中的物体进行快速且准确的目标识别是提升机器人智能度的重要突破口.针对家庭服务机器人竞赛,提出一种目标识别算法,使用轻量级网络MobileNetv2替换YOLOv5的主干特征提取网络,实验证明该识别算法在快速性和准确性方面有一定的提升.     

基于树莓派与YOLOv5-Lite模型的行人检测系统设计

把在PC端上训练好的YOLOv5s与YOLOv5-Lite目标检测模型分别部署在搭载Linux系统的树莓派4B平台上,并在此平台上搭建深度学习环境,构建道路行人检测系统。对这两个模型进行分析对比,实验结果表明,在识别准确率相差0.1%的情况下,YOLOv5-Lite模型相对于原YOLOv5s模型,网络参数量下降了78.26%,模型计算量下降了77.91%,模型内存大小下降了75.52%,检测速度提高了91.67%。综上,本文提出的基于树莓派和轻量化YOLOv5-Lite目标检测网络模型的行人检测系统兼顾了识别准确、适用性好、小型化、成本低等综合性能优势。     

基于边缘设备和改进YOLOv5算法的车牌号码识别

自动识别车牌号码是智慧交通中的重要内容,针对现有车牌识别算法计算量大,不满足微型化、实时性等需求,提出一种基于边缘设备和改进YOLOv5算法的车牌号码识别方法。首先,构建车牌数据集;其次,通过改进YOLOv5网络模型架构,并引入注意力机制,提升对车牌号码的检测能力,并与未改进的YOLOv5算法作性能对比;最后,将Intel Movidius NCS2与树莓派硬件设备结合,进行实时推理。实验结果表明,改进的YOLOv5算法在边缘设备上的实时画面推理速度最快达到3.316 ms,YOLOv5算法推理速度为5.772 ms,改进的YOLOv5算法与原算法相比,其推理速度平均提升了13.41%。本文提出的方法能在边缘设备上提高车牌检测速度,并达到较高的准确率。     

基于STM32的武术擂台机器人的设计

介绍以STM32单片机为控制器的武术擂台机器人的设计,以及在研究与实践过程中遇到的问题及解决方案.     

一种小型履带式管道机器人的研究与设计

针对通风管道人工探查不便的现状,设计了一款专用的小型履带机器人。该机器人通过合理的结构设计,可以在狭窄的管道内实现爬坡和越障功能。开发了基于树莓派和STM32的控制系统,设计了整个机器体的硬件电路和各模块的控制程序,利用Qt Creator软件创建了便于控制与接收数据的上位机界面。通过光纤通信实现上位机和机器人的数据互通。机器人搭载的摄像头模块和轮速计可以在复杂的管道环境中实现定位功能。经过实验与数据采集,表明该机器人整体功能完善,在实际工作环境中运行平稳,能够满足管道机器人的工作要求。     

基于YOLOv5的交通标志识别

交通标志识别是自动驾驶技术中的关键一部分.针对交通标志在道路场景中目标较小且识别精度较低的问题,提出一种改进的YOLOv5算法.首先在YOLOv5模型中引入全局注意力机制(GAM),提高网络捕获不同尺度交通标志特征的能力;其次将YOLOv5算法中使用的GIoU损失函数更换为更具回归特性的CIoU损失函数来优化模型,提高对交通标志的识别精度.最后在Tsinghua-Tencent 100K数据集上进行训练,实验结果表明,改进后的YOLOv5算法对交通标志识别的平均精度均值为93.00%,相比于原算法提升了5.72%,具有更好的识别性能.     

基于YOLOv5算法的钢轨扣件分类检测研究

钢轨扣件是保障轨道车辆安全运行的基础,目前依靠人工检查钢轨扣件状态效率低且具有缺陷性。基于YOLOv5算法对钢轨弹条断裂、缺失、移位以及螺栓缺失四种状态扣件进行分类检测研究,文章选取706张含故障扣件的图片进行标注形成钢轨故障扣件数据集,讨论了YOLOv5s、YOLOv5m两种模型对数据集分别训练50、100次后的识别效果,结果显示：采用YOLOv5m模型训练100次的YOLOv5算法对各类别故障扣件的测试精度、召回率、mAP@.5、mAP@.5:.95分别为0.988、0.967、0.987、0.822。该方法对钢轨扣件分类检测具有很好的应用价值。     

ROBOT_S: An interactive robot simulation language

While it may not be practical to realize a tentative robot design as an actual robot, there is no question of the practicality of a simulation, ROBOT_S is a program in which the foundation for a comprehensive simulation environment is laid. A graphical robot is created to which physical attributes may be assigned, and whose movement may be dictated by a user-installed dynamic model and control law. A simple robot command language has been developed, by which the manipulator may be commanded to move, during which simulation data of state variables is collected and graphed.     

基于优化YOLOv5s的跌倒人物目标检测方法

针对目标检测模型在人物跌倒时易漏检、鲁棒性和泛化能力差等问题,提出一种基于改进YOLOv5s的跌倒人物目标检测方法 YOLOv5s-FPD。首先,对Le2i跌倒数据集使用多种方式扩充后用于模型训练,增强模型鲁棒性和泛化能力;其次,使用MobileNetV3作为主干网络来进行特征提取,协调并平衡模型的轻量化和准确性关系;然后,利用BiFPN改善模型多尺度特征融合能力,提高了融合速度和效率,并使用CBAM轻量级注意力机制实现注意力对通道和空间的双重关注,增强了注意力机制对模型准确性地提升效果;最后,引入Focal Loss损失评价从而更注重挖掘困难样本特征,改善正负样本失衡的问题。实验结果表明,在Le2i跌倒数据集上YOLOv5s-FPD模型比原YOLOv5s模型,在精确度、F₁分数、检测速度分别提高了2.91%,0.03和8.7 FPS,验证了该方法的有效性。     

基于预测的足球机器人截球动作的研究

本文提出了一种改进的足球机器人截球动作的算法,该算法给出了球在摩擦力作用下的运动模型并基于线性回归预测模型对其在场地上的运动作了分析,结合足球机器人的到定点的运动模型进而使用二分法求解拦截,实现了截球动作.实验表明该算法具有良好的拦截效果.     

An interactive tool for mobile robot motion planning

This paper presents an interactive tool aimed at facilitating the understanding of several well-known algorithms and techniques involved in solving mobile robot motion problems. These range from those modelling the mechanics of mobility to those used in navigation. The tool focuses on describing these problems in a simple manner in order to be useful for education purposes among different disciplines. By highlighting interactivity, the tool provides a novel means to study robot motion planning ideas in a manner that enhances full understanding. Furthermore, the paper discuses how the tool can be used in an introductory course of mobile robotics.     

基于改进YOLOv5的安全绳目标检测

在工业施工过程中, 工人安全已成为一个日益重要的问题, 佩戴安全绳等安全装备是保护工人在高处工作时生命安全的重要措施;在现代化生产施工过程中, 通过使用监控摄像设备结合人工智能算法的方式来检测工人佩戴安全绳等设备越发普遍, 但安全绳由于细长、形状多变以及环境变化等因素较为难以准确识别;为解决以上问题, 并确保能够在不同环境下能够准确识别安全绳, 现提出一种使用YOLOv5目标检测算法, 首先通过改进的FasterNet模块进行上下文信息提取, 在Neck网络中使用改进的多维动态卷积保留更多特征信息, 使用WIoU_Loss损失函数来提高定位精度, 在训练过程中使用动态调整学习率的策略;实验结果表明, 改进后的算法在降低计算复杂度的情况下提高了3.0%的检测精度, mAP@0.5提高了4.3%, 经过在实际场景应用, 满足项目对实时检测精度及速度的要求。     

An insectomorphic robot climbing over a freely rolling ball

An algorithm for the control of an insectomorphic robot climbing over a ball that rolls freely on a horizontal plane is developed and tested using computer simulation. The proposed motion involves three maneuvers. First, the robot climbs the ball at rest from the horizontal surface. At the end of this maneuver, the ball gains an angular velocity due to errors in the execution of the programmed motion. The further motion of the robot is designed so as to reduce the velocity gained in the course of climbing to an acceptable level. The motion is completed by the maneuver of getting down to the supporting horizontal plane from the almost motionless ball. The robot motion is implemented using the Coulomb friction without any special devices. The asymptotic stability of the programmed motion of the system as a whole is ensured by a PD controller that implements the step cycles of the leg motions and the planned motion of the body. Results of 3D computer simulation of the robot motion are discussed. The model of the mechanical robot-ball system is formed using the Universal Mechanism program package; this model is described by an automatically derived system of differential equations that take into account the dynamics of all solid elements.     

基于改进YOLOv5的桥梁裂缝模型研究

桥梁裂缝人工检测耗时费力、安全性不高,为了高效、准确、无接触地对桥梁裂缝进行识别检测,提出一种基于改进YOLOv5的桥梁裂缝检测模型YOLOv5-SA;该方法在YOLOv5s模型的基础上,首先对收集的数据集利用几何变换、光学变换等操作进行数据增强;其次将融合视觉注意力机制(SKNet)添加到Head部分来提高模型对裂缝特征的表示能力;最后在金字塔特征表示法(FPN)的基础上利用自适应空间特征融合(ASFF)模块加强网络特征融合能力,增加对桥梁裂缝小目标的检测;结果表明：改进后的模型相对于YOLOv5s模型能更好地抑制非关键信息,减少背景中的无效信息干扰,提高桥梁裂缝目标检测精准度;改进后的YOLOv5-SA模型准确率达到88.1%,与原YOLOv5s模型相比提高了1.6%;平均精度均值mAP 0.5和mAP 0.5～0.95分别达到90.0%、62.1%,相比而言分别提高了2.2%、2.4%;与其他桥梁裂缝检测相关方法(Faster-RCNN、YOLOv4tiny)相比,提出的YOLOv5-SA模型也具有相当或更好的检测性能;由此可见改进后的模型能更高效地检测复杂环境下的桥梁裂缝,可以...     

Multiple fuzzy state-value functions for human evaluation through interactive trajectory planning of a partner robot

The purpose of this study is to develop partner robots that can obtain and accumulate human-friendly behaviors. To achieve this purpose, the entire architecture of the robot is designed, based on a concept of structured learning which emphasizes the importance of interactive learning of several modules through interaction with its environment. This paper deals with a trajectory planning method for generating hand-to-hand behaviors of a partner robot by using multiple fuzzy state-value functions, a self-organizing map, and an interactive genetic algorithm. A trajectory for the behavior is generated by an interactive genetic algorithm using human evaluation. In order to reduce human load, human evaluation is estimated by using the fuzzy state-value function. Furthermore, to cope with various situations, a self-organizing map is used for clustering a given task dependent on a human hand position. And then, a fuzzy state-value function is assigned to each output unit of the self-organizing map. The robot can easily obtain and accumulate human-friendly trajectories using a fuzzy state-value function and a knowledge database corresponding to the unit selected in the self-organizing map. Finally, multiple fuzzy state-value functions can estimate a human evaluation model for the hand-to-hand behaviors. Several experimental results show the effectiveness of the proposed method.     

An interactive programming system for the IBM 7545 robot

Industrial robots may be programmed using teach methods, off-line programming languages or by using interactive robot programming systems. This paper briefly explains each method, describes the advantages of developing interactive robot programming systems, and then describes an interactive robot programming system developed for the IBM 7545 robot. The approach used in the development process, the interactive execution and user options, and a demonstration of the operation of this interactive robot programming system are also presented.     

Computational intelligence for structured learning of a partner robot based on imitation

Imitation is a powerful tool for gestural interaction between children and for teaching behaviors to children by parent. Furthermore, others’ action can be a hint for acquiring a new behavior that might not be the same as the original action. The importance is how to map or represent others’ action into new one in the internal state space. A good instructor can teach an action to a learner by understanding the mapping or imitating method of the learner. This indicates a robot also can acquire various behaviors using interactive learning based on imitation. This paper proposes structured learning for a partner robot based on the interactive teaching mechanism. The proposed method is composed of a spiking neural network, self-organizing map, steady-state genetic algorithm, and softmax action selection. Furthermore, we discuss the interactive learning of a human and a partner robot based on the proposed method through experiment results.