Image‐based visual target following for a nonholonomic wheeled mobile robot with a single fixed camera

This paper addresses the problem of visual target following for a nonholonomic wheeled mobile robot with a single fixed camera. For the problem, we propose a control scheme composed of vision‐based trajectory planning and tracking control. We present online trajectory planning based on image‐based visual servoing. The key idea is to compensate the target object motion. In particular, we focus on visual feature motion resulting from the target object motion. Tracking control to the generated trajectory can therefore achieve visual target following precisely even if the target object is moving. Our proposed scheme can be applied to, e.g. robotic systems for autonomous guard and observation. The effectiveness of the control system is evaluated experimentally. © 2012 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.     

Sonar-Based Rover Navigation for Single or Multiple Platforms: Forward Safe Path and Target Switching Approach

In this paper, we have proposed a sensor fusion scheme along with the geometrical modeling of mobile robot navigation path in an unknown environment. In this scheme, the physical placement of sonars, their ranging limits and beam opening angles are considered. A simple 2-D axis transformation is proposed to relate local robot frame with the actual navigation environment. forward safe path (FSP) and target switching approach (TSA) are proposed for efficient obstacle avoidance and target tracking of mobile robot. FSP greatly simplifies the environment conditions as sensed by the robot and also provides minimum turning path during avoidance of obstacles. This method also removes the ldquooscillationrdquo in the mobile robot navigation path. TSA technique gives highest priority on the target tracking during the obstacle avoidance and seeks minimum distance path towards the target. These methods remove unnecessary turning of mobile robot during navigation. A scheme for target directional motion is also proposed. So, mobile robot takes the minimum turning path required towards the target. These methods also ensure the avoidance of ldquodead cycle problemrdquo. These schemes are successfully implemented on a model of PatrolBot mobile robot from ActivMedia Robotics. The overview of current research work on multi-domain robotic system namely system-of-systems is also presented. This paper also describes the Global Positioning System-based navigation of rovers. Results of real-time experiments with Pioneer II P2AT-8 from ActivMedia are included in this paper to show the future aspect of this research work.     

融合行人运动信息的室内移动机器人动态避障方法

为了提高移动机器人在室内人机共融环境下的运动安全和交互性,提出了一种融合行人运动信息的室内移动机器人动 态避障方法,同时考虑任务约束和社会规则。 首先,利用 YOLO v3 算法和 Deep Sort 算法分别对室内环境中的行人进行实时检 测与目标跟踪,计算行人在过去时刻的历史轨迹。 然后,利用 Social-GAN 算法构建行人交互模型,实现轨迹预测。 在此基础上, 将行人的运动状态融合进机器人避障算法之中,根据社会规则设计评价函数,对机器人采样速度样本进行评估,使移动机器人 能够以安全和舒适的方式绕过行人,确保室内人机共融环境下移动机器人的社会接受性。 通过实验对比分析,与传统 DWA 方 法相比,本文方法不仅可以提高机器人导航避障效率,在相同室内场景下导航避障时间由 23. 56 s 提高到 19. 38 s,而且可以有 效降低与行人发生碰撞的风险,保证机器人导航的安全和社交性。     

基于ARM-Linux的爬壁机器人控制器研究

针对人工船舶除锈的效率低、危险性高、成本高等问题,研制出爬壁机器人进行船舶除锈。采用了上、下位机的分布式控制方案,对基于ARM-Linux的爬壁机器人控制器进行研究,包括遥感器、比较器及其组成电路。通过引用强化学习算法Q-Leaning算法,实现了爬壁机器人的强化学习循迹,改进了传统PID等算法无法针对环境进行最优化动作策略选择的缺点,提高爬壁机器人在不同环境下循迹的准确性。实验结果表明,基于ARM-Linux的爬壁机器人控制系统性能较好,可以满足控制器要求。     

Guiding a mobile robot with cellular neural networks

We show how cellular neural networks (CNNs) are capable of providing the necessary signal processing needed for visual navigation of an autonomous mobile robot. In this way, even complex feature detection and object recognition can be obtained in real time by analogue hardware, making fully autonomous real‐time operation feasible. An autonomous robot was first simulated and then implemented by simulating the CNN with a DSP. The robot is capable of navigating in a maze following lines painted on the floor. Images are processed entirely by a CNN‐based algorithm, and navigation is controlled by a fuzzy‐rule‐based algorithm. Copyright © 2002 John Wiley & Sons, Ltd.     

基于神经网络视觉伺服的机器人模糊自适应阻抗控制

提出一种未知环境下基于神经网络视觉伺服的机器人模糊自适应阻抗控制策略.首先使用BP神经网络来学习曲线图像特征变化率与机器人关节角速度的映射关系,其次由机器人力伺服控制得到离散阻抗控制模型,并根据接触力的变化对阻抗模型参数进行模糊调节,减少了受限运动中力干扰的影响.最后通过末端固定安装单摄像机和力传感器的6自由度机器人跟...     

变电站设备巡检机器人视觉导航方法

针对变电站巡检工作的实际需求,介绍了一种基于轨线引导的巡检机器人单目视觉导航方法。对于预先设置的引导轨线和停靠位标识,采用OpenCV和DirectShow软件库来开发视频采集程序,实现道路视频信息的实时采集;通过HSI模型对图像进行颜色提取,再采用数学形态学处理去除干扰信息;最终采用统计8-连通区的方法将导航线和停靠位标识分割提取出来,用采样估算方法计算导航线参数,通过比例系数融合导航参数以控制两驱动轮电机实现机器人的运动控制。实验表明,该方法实现了机器人的自主导航,能够适应绝大部分光照强度的环境,并具有较好的实时性和可靠性。     

悬停状态下旋翼飞行机器人自适应鲁棒控制

针对加装二自由度绳驱动机械臂的旋翼飞行机器人在悬停条件下的抗干扰控制,提出了一种自适应终端滑模控制策略.将系统分成四旋翼飞行器和机械臂两个子系统,分别采用拉格朗日法与牛顿-欧拉法获得各自的动力学模型.在Lyapunov稳定性框架下设计了旋翼飞行器人的抗干扰轨迹跟踪控制器,并引人自适应策略来估计扰动上界.通过3个仿真算例验证了所设计控制器的有效性,结果表明,与其他控制器相比,本控制器具有较高的跟踪精度、较强的鲁棒性以及较好的抗干扰能力;机械臂质量的变化主要影响x通道和y通道的控制性能;本控制器基本能满足旋翼飞行机器人悬停作业的工作需求,具有一定的工程参考意义.     

Non-linear learning control of robot manipulators without requiring acceleration measurement

A new class of non-linear learning control laws is introduced for a robot manipulator to track a given trajectory in performing a series of tasks. The learning control scheme is applicable to robots with both resolute and prismatic joints, requires only position and velocity feedback, and removes the acceleration measurement required by the existing results. It has been shown that under the proposed learning control the tracking errors are always guaranteed to be asymptotically stable with respect to the number of trials. The proposed control is robust in the sense that exact knowledge about the non-linear dynamics is not required except for bounding functions on the magnitudes. In addition, the new learning scheme can be used without assumptions such as repeatability of robot motion, repeatability of tasks and resetting of initial tracking errors.     

Control of biped walking robot for human living environment

This paper introduces biped robot adaptation to human living environment from viewpoints of battery operation time extension and environmental recognition. These issues are important when robots actually work at home. First, in order to extend battery operation time, we propose energy-saving bipedal locomotion gait. The problem is formulated as an optimal control problem, which is conventionally hard to solve when a target system is complicated. In this paper, partial derivatives appeared in optimal control problem are implicitly represented by using automatic differentiation technique. This approach enables complicated optimal control problem solvable. In combination with receding horizon control, its computation cost is also reduced. Second, we introduce the biped walk tracking based on the camera image mounted on the walking robot, and the visual servoing by the posture change for the purpose of the target image tracking in the camera frame. We propose a new control law to track the rotated target object using the characteristic of the walking, which considered the interference between translational motion and rotational motion. The decoupling is realized by simulations and experiments. As a result, the walking robot tracked the translated and rotated target object without a practical issue. Copyright © 2009 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.     

基于深度信息的巡逻机器人避障系统实现

针对巡逻机器人的避障导航,基于Realsense深度摄像头获取环境深度信息,在使用机载计算机对深度图像进行分层处理的基础上,设计了巡逻机器人避障导航系统,同时对巡逻机器人通信协议以及决策流程等进行了改进设计,最后在ROS机器人开源操作系统上对其进行实验及验证。实验结果表明,在相同环境下,该方法能使巡逻机器人快速、安全、有效地避开静态障碍物而到达目标点。该方法不仅能够提高避障的成功率,获得实时避障效果,而且能够降低巡逻机器人的数据处理量,有利于巡逻机器人的移植。     

Motion of control of a multijoint robot based on a robust velocity controller

This paper proposes a simple and robust robot motion control method using a robust velocity controller. The robust velocity controller is based on H^∞ control theory, and is called H^∞ velocity controller. The H^∞ velocity controller based motion control method is completely equivalent to the robust acceleration control method using the H^∞ acceleration controller, but it has simpler structure. Therefore, the proposed system can realize a fine robot motion control easily. To confirm the validity of the proposed method, this paper realizes the hybrid control of position and force for a multijoint robot manipulator. Further, the simple realization of hybrid control is proposed considering the attitude of the robot manipulator. This system achieves hybrid control of position and force of the robot manipulator while maintaining a perpendicular attitude to the target environment. The experimental results in this paper show that the proposed system has the desired force and position response to the target environment. © 1997 Scripta Technica, Inc. Electr Eng Jpn, 118 (4): 58–69, 1997     

结合拆分注意力机制和下一次预期观察的视觉导航

针对深度强化学习视觉导航算法因导航场景变化而导致导航精度下降,影像匹配的实时性和可靠性降低的问题,提出一种融合拆分注意力机制和下一次预期观测(NEO)的视觉导航模型。首先使用ResNest50骨干网络提取当前状态和目标状态的特征以降低网络冗余,利用跨阶段部分连接CSP强化捕获浅层目标特征信息以增强模型的学习能力。然后提出改进的损失函数,使得推理网络更加接近于真实后验,从而智能体能在当前环境下做出最佳决策,进一步提升不同场景下模型的导航精度。在AVD数据集和AI2-THOR场景进行训练及测试,实验结果表明,本文算法导航精度高达96.8%,平均SR提升约3%,平均SPL提升约6%,可以满足导航精度和实时匹配的要求。     

Autonomous mobile robot based on behavior decision skill and control skill of the operator

This paper introduces a human skill base control algorithm using artificial neural networks and fuzzy reasoning for an autonomous mobile robot. Neural networks are used to select a suitable motion control pattern in actual environments. The back propagation algorithm adjusts the weights of the neural networks so that the selected motion control pattern corresponds to the action, which is obtained by the operator's behavior decision skill. To realize the selected motion control pattern, the orientation angle and the speed of the mobile robot are determined by fuzzy reasoning in which fuzzy rules are also automatically tuned so as to simulate the operator's control skill. We have implemented and tested the proposed control algorithm on an autonomous mobile robot and some experimental results demonstrate the effectiveness of the proposed control algorithm for the autonomous mobile robot. © 2000 Scripta Technica, Electr Eng Jpn, 131(2): 30–39, 2000     

Autonomous Navigation of Vehicles from a Visual Memory Using a Generic Camera Model

In this paper, we present a complete framework for autonomous vehicle navigation using a single camera and natural landmarks. When navigating in an unknown environment for the first time, usual behavior consists of memorizing some key views along the performed path to use these references as checkpoints for future navigation missions. The navigation framework for the wheeled vehicles presented in this paper is based on this assumption. During a human-guided learning step, the vehicle performs paths that are sampled and stored as a set of ordered key images, as acquired by an embedded camera. The visual paths are topologically organized, providing a visual memory of the environment. Given an image of the visual memory as a target, the vehicle navigation mission is defined as a concatenation of visual path subsets called visual routes. When autonomously running, the control guides the vehicle along the reference visual route without explicitly planning any trajectory. The control consists of a vision-based control law that is adapted to the nonholonomic constraint. Our navigation framework has been designed for a generic class of cameras (including conventional, catadioptric, and fisheye cameras). Experiments with an urban electric vehicle navigating in an outdoor environment have been carried out with a fisheye camera along a 750-m-long trajectory. Results validate our approach.     

Robot vision with cellular neural networks: a practical implementation of new algorithms

Cellular neural networks (CNNs) are well suited for image processing due to the possibility of a parallel computation. In this paper, we present two algorithms for tracking and obstacle avoidance using CNNs. Furthermore, we show the implementation of an autonomous robot guided using only real‐time visual feedback; the image processing is performed entirely by a CNN system embedded in a digital signal processor (DSP). We successfully tested the two algorithms on this robot. Copyright © 2006 John Wiley & Sons, Ltd.     

基于图像雅可比矩阵的机器人视觉/力控制

在实际的接触型作业任务中,需要在控制位置的同时,控制末端执行器与环境之间的接触力。针对这一问题,首先通过力反馈信息对未知环境中的法线方向进行估计,然后设计了一种基于模糊自适应Kalman滤波的动态图像雅可比矩阵辨识方法,并将其应用于机器人视觉反馈控制任务中。最后提出视觉/力混合控制算法,在该算法中给出了视觉/力混合控制的约束条件,对视觉采用变结构控制器,力控制采用PI控制器。仿真结果表明,该策略具有较高的力控制精度和曲线跟踪能力。     

变电站二次设备巡检的Mecanum轮式机器人精确定位与作业辨识

通过巡检机器人进行变电站二次设备的监测是提升电力设备自动化、智能化管理的重要方式,有利于保障电力工程设备的安全运行。本研究开发了一种用于变电站二次设备自动化巡检的Mecanum轮式机器人,具备自主导航定位与作业辨识的能力,可极大提升设备巡检效率及保护压板状态识别准确性。通过Mecanum轮的驱动方式实现巡检机器人在狭窄作业环境下的灵活运动与姿态调整,多轨道升降平台实现对350-1800 mm高度范围内的二次设备及压板的图像采集与状态辨识。机器人采用基于激光雷达的SLAM导航方法进行自主定位导航,并结合基于视觉的路径提取与跟踪算法进行姿态位置修正,实现在待测点位置的精确定位。同时,提出了基于颜色辨识的图像排列与状态辨识方法,针对二次设备保护压板连通状态进行识别和判断。实验结果表明,研制的变电站二次设备巡检机器人可以实现自主导航与位置精确定位,在路径跟踪过程中最大偏角和偏距分别为±3°和±8 mm。结合机器视觉与颜色辨识的压板辨识方法可以准确识别压板状态,识别准确率大于95.80%,有助于提升机器人自动化的电力巡检作业水平。     

Adaptive predictive control of a differential drive robot tuned with reinforcement learning

One of the most important steps in designing a model predictive control strategy is selecting appropriate parameters for the relative weights of the objective function. Typically, these are selected through trial and error to meet the desired performance. In this paper, a reinforcement learning technique called learning automata is used to select appropriate parameters for the controller of a differential drive robot through a simulation process. Results of the simulation show that the parameters always converge, although to different values. A controller chosen by the learning process is then ported to a real platform. The selected controller is shown to control the robot better than a standard model predictive control.     

Autonomous driving control for ITS by using learning automaton network with multiple environments

In this paper, the application of the learning automaton (LA) network with multiple environments is proposed for the adaptive controller for ITS autonomous driving. The LA network, which we introduced previously, has the ability to learn which deals with both multiple reinforcement signals and information of multiple environments at the same time. This feature is found to be useful for improving the response of adaptation in dynamic environments such as highways. In order to evaluate the practical advantage of using the network, we designed a simulational highway system, constructed an autonomous travel controller using the simple LA and the LA network, and executed comparative experiments evaluating the performance of adaptation response and collision avoidance. The results show that the performance of the LA network with multiple environments is superior to that using simple LA application with regard to stability and safety. © 2005 Wiley Periodicals, Inc. Electr Eng Jpn, 150(4): 36–43, 2005; Published online in Wiley InterScience ( www.interscience.wiley.com ). DOI 10.1002/eej.10363