Decentralized Traffic Control for Non-Holonomic Flexible Automated Guided Vehicles in Industrial Environments

This paper deals with the problem of coordinating flexible automated guided vehicles (AGVs) in real manufacturing systems. The problem consists of ensuring safe and successful task execution while several AGVs operate as a distributed transportation system in real industrial environments. The proposed solution combines different decentralized techniques to increase the flexibility and scalability of the multirobot system. The coordination is addressed by dividing the problem into path planning, obstacle avoidance and traffic control problems. The path planning method takes into account the location of mates for replanning the routes. The obstacle avoidance technique considers the kinematic constraints of the platform for reactive motion control. The traffic control approach makes use of a decentralized control policy that takes into account the capabilities of vehicles. By combining all these techniques and configuring the system properly, we present the successful development of a distributed transportation system composed of a team of flexible AGVs. The proposed solution has been validated using both a set of custom-modified AGVs operating in a real factory and a simulation of several AGVs operating in a virtual scenario.     

基于改进蚁群优化算法的自动驾驶多车协同运动规划

当前面向多辆自动驾驶汽车的协同运动规划方法能有效保证运行车辆与障碍物及其他车辆之间避免发生碰撞并保持安全距离,但车辆间的在线协同与规划能力仍有待提升。为实现多辆自动驾驶汽车在运动过程中的协同控制,提出一种基于改进蚁群优化算法的多车在线协同规划方法。以空间协同与轨迹代价为优化目标,构造多目标优化函数,确保了多车行驶过程中的协同安全性与轨迹平滑性。将多目标优化函数引入蚁群优化算法的信息素更新过程中,根据自动驾驶车辆数量产生多个种群,使得种群之间相互独立的同时为每辆自动驾驶汽车规划可行路线。最终对蚁群优化算法中的挥发因子进行自适应调整,提升了算法全局搜索能力及收敛速度。实验结果表明,该方法能使多辆自动驾驶汽车在运动过程中保持协同控制并规划出无碰撞路线,相比于基于人工势场和模型预测的协同驾驶方法在复杂道路场景下车辆间的协同效果更好且适应性更强。     

Coordinated search for a lost target in a Bayesian world

This paper describes a decentralized Bayesian approach to the problem of coordinating multiple autonomous sensor platforms searching for a single non-evading target. In this architecture, each decision maker builds an equivalent representation of the probability density function (PDF) of the target state through a general decentralized Bayesian sensor network, enabling them to coordinate their actions without exchanging any information about their plans. The advantage of the approach is that a high degree of scalability and real-time adaptability can be achieved. The framework is implemented on a real-time high-fidelity multi-vehicle simulator system. The effectiveness of the method is demonstrated in different scenarios for a team of airborne search vehicles looking for both a stationary and a drifting target lost at sea.     

Multi-vehicle consensus with a time-varying reference state

In this paper, we study the consensus problem in multi-vehicle systems, where the information states of all vehicles approach a time-varying reference state under the condition that only a portion of the vehicles (e.g., the unique team leader) have access to the reference state and the portion of the vehicles might not have a directed path to all of the other vehicles in the team. We first analyze a consensus algorithm with a constant reference state using graph theoretical tools. We then propose consensus algorithms with a time-varying reference state and show necessary and sufficient conditions under which consensus is reached on the time-varying reference state. The time-varying reference state can be an exogenous signal or evolve according to a nonlinear model. These consensus algorithms are also extended to achieve relative state deviations among the vehicles. An application example to multi-vehicle formation control is given as a proof of concept.     

Stabilization of sets with application to multi-vehicle coordinated motion

In this paper, we develop stability and control design framework for time-varying and time-invariant sets of nonlinear dynamical systems using vector Lyapunov functions. Several Lyapunov functions arise naturally in multi-agent systems, where each agent can be associated with a generalized energy function which further becomes a component of a vector Lyapunov function. We apply the developed control framework to the problem of multi-vehicle coordinated motion to design distributed controllers for individual vehicles moving in a specified formation. The main idea of our approach is that a moving formation of vehicles can be characterized by a time-varying set in the state space, and hence, the problem of distributed control design for multi-vehicle coordinated motion is equivalent to the design of stabilizing controllers for time-varying sets of nonlinear dynamical systems. The control framework is shown to ensure global exponential stabilization of multi-vehicle formations. Finally, we implement the feedback stabilizing controllers for time-invariant sets to achieve global exponential stabilization of static formations of multiple vehicles.     

Scalable and Efficient Video Coding Using 3-D Modeling

In this paper, we present a three-dimensional (3D) model-based video coding scheme for streaming static scene video in a compact way but also enabling time and spatial scalability according to network or terminal capability and providing 3D functionalities. The proposed format is based on encoding the sequence of reconstructed models using second-generation wavelets, and efficiently multiplexing the resulting geometric, topological, texture, and camera motion binary representations. The wavelets decomposition can be adaptive in order to fit to images and scene contents. To ensure time scalability, this representation is based on a common connectivity for all 3D models, which also allows straightforward morphing between successive models ensuring visual continuity at no additional cost. The method proves to be better than previous methods for video encoding of static scenes, even better than state-of-the-art video coders such as H264 (also known as MPEG AVC). Another application of our approach are smoothing camera path for suppression of jitter from hand-held acquisition and the fast transmission and real-time visualization of virtual environments obtained by video capture, for virtual or augmented reality and interactive walk-through in photo-realistic 3D environments around the original camera path     

Experimental validation of collective circular motion for nonholonomic multi-vehicle systems

This paper presents the experimental validation of a recently proposed decentralized control law, for the collective circular motion of a team of nonholonomic vehicles about a virtual reference beacon. The considered control strategy ensures global asymptotic stability in the single-vehicle case and local asymptotic stability in the multi-vehicle scenario. The main contribution of this work is to evaluate the performance of the proposed algorithm in the presence of a number of uncertainty sources naturally arising in a real-world environment. Both static and moving reference beacons are considered, in a low-cost experimental framework based on the LEGO MINDSTORMS technology. The adopted setup features good scalability and is versatile enough to be adopted for the evaluation of different control strategies. At the same time, it represents a challenging testbed, exhibiting several issues that have to be faced in real-world applications.     

Distributed output feedback stationary consensus of multi-vehicle systems in unknown environments

In the traditional distributed consensus of multi-vehicle systems, vehicles agree on velocity and position using limited information exchange in their local neighborhoods. Recently, distributed leaderless stationary consensus has been proposed in which vehicles agree on a position and come to a stop. The proposed stationary consensus schemes are based on all vehicles'' access to their own absolute velocity measurements, and they do not guarantee this collective behavior in the presence of disturbances that persistently excite vehicles'' dynamics. On the other hand, traditional distributed disturbance rejection leaderless consensus algorithms may result in an uncontrolled increase in the speed of multi-vehicle system. In this paper, we propose a dynamic relative-output feedback leaderless stationary algorithm in which only a few vehicles have access to their absolute measurements. We systematically design the distributed algorithm by transforming this problem into a static feedback robust control design challenge for the low-order modified model of vehicles with fictitious modeling uncertainties. We further propose dynamic leader-follower stationary consensus algorithms for multi-vehicle systems with a static leader, and find closed-form solutions for the consensus gains based on design matrices and communication graph topology. Finally, we verify the feasibility of these ideas through simulation studies.     

基于残差网络的小型车辆目标检测算法

城市道路中车辆检测与识别对于提升交通安全,发展智能化交通具备非常重要的意义。传统的检测方式依赖于人工提取的特征,已难以适用于复杂多变的交通场景,存在识别精确度低、时间复杂度高等缺陷。深度学习模型可以自动提取有用特征,泛化能力强,但难以对相似型车辆进行更加精细的分类,为此提出一种基于残差网络的小型车辆目标检测算法。算法将传统卷积神经网络的连接形式改为一种基于局部连接和权值共享的残差连接模式,同时更改网络结构控制参数数量,将图片不同层次的特征融合计算,应用感兴趣区域池化层规格化前层特征,最后经过分类层和回归层得到目标框的置信度以及修正参数。实验表明,改进模型能够在保证时间效率的前提下增强网络的学习能力,提高平均精度,在相似小型车辆的检测问题上取得了良好的检测结果。     

典型交通场景下智能车的定位与协作

提出并设计了基于多智能体的多智能车协作系统平台,包括基于嵌入式多车协作系统的硬件、软件及通信协议,该系统可作为多车协作算法的仿真实验平台。针对典型城市交通场景,提出了基于拓扑结构的两层地图模型,并采用视觉与里程结合的方法实现智能车在地图中的准确定位。在此基础上,设计了基于冲突表的路口协调算法、车队跟随算法,并完成路口、车队、超车等协作任务,通过实验结果证明了多车协作系统定位与协作的可行性和可靠性。     

Collective circular motion of multi-vehicle systems

This paper addresses a collective motion problem for a multi-agent system composed of nonholonomic vehicles. The aim of the vehicles is to achieve circular motion around a virtual reference beacon. A control law is proposed, which guarantees global asymptotic stability of the circular motion with a prescribed direction of rotation, in the case of a single vehicle. Equilibrium configurations of the multi-vehicle system are studied and sufficient conditions for their local stability are given, in terms of the control law design parameters. Practical issues related to sensory limitations are taken into account. The transient behavior of the multi-vehicle system is analyzed via numerical simulations.     

基于凸近似的避障原理及无人驾驶车辆路径规划模型预测算法

提出了一种改进的无人驾驶车辆路径规划方法,并搭建了相应的软件在环实时仿真系统,对方法在结构化道路下的复杂动态交通工况进行性能验证.首先,引入基于凸近似的避障原理,对障碍物参考点的选取进行优化,扩大了路径规划的可行域范围.采用改进后的方法,并结合模型预测控制(Model predictive control,MPC)原理和曲线坐标系统,综合考虑自车及障碍车的外形、道路几何约束及自车的机械结构约束、路径最短、侧向加速度、道路对中、逐次变道、车距安全度、左变道优先和前轮转角变化率等权重的影响,实现了车辆在复杂动态交通工况下的路径规划.最后,以长城H7运动多用途车作为无人驾驶实验及仿真平台,搭建基于dSPACE多核架构的Carsim+Simulink软件在环实时仿真系统,对算法进行验证.结果表明,所提出的方法不仅可获得合理、平滑的行驶路径,顺利避开运动障碍车的干扰,而且具有良好的实时性.     

Safe Reinforcement Learning Algorithm and Its Application in Intelligent Control for CPS

The design of a safe controller for a Cyber-Physical System (CPS) is a hot research topic. The existing safety controller design based on formal methods has problems such as excessive reliance on models and poor scalability. Intelligent control based on Deep Reinforcement Learning (DRL) can handle high-dimensional nonlinear complex systems and uncertain systems and is becoming a very promising CPS control technology, but it lacks safety guarantees. This study addresses the safety issues of Reinforcement Learning (RL) control by analyzing a typical case of an industrial oil pump control system and carries out research on a Safe Reinforcement Learning (SRL) algorithm and intelligent control application. First, the SRL issue of the industrial oil pump control system is formalized, and a simulation environment of the oil pump is built. Then, by designing the structure and activation function of the output layer, an oil pump controller in the form of a neural network is constructed to satisfy the linear inequality constraints of the on-off operations of the oil pump. Finally, in order to better balance the safety and optimality control objectives, a new SRL algorithm is designed based on the Augmented Lagrange Multiplier (ALM) method. A comparative experiment on the industrial oil pump shows that the controller synthesized by the proposed algorithm surpasses existing similar algorithms both in safety and optimality. During the evaluation, the neural network controllers synthesized in this study pass rigorous formal verification with a probability of 90%. Meanwhile, compared with the theoretically optimal controller, neural network controllers achieve an optimal objective value loss of 2%. The proposed method is expected to be applied in more scenarios, and the case study scheme may provide a reference for other researchers in the field of safe intelligent control and formal verification.     

Decentralized control of a group of mobile robots for deployment in sweep coverage

This paper addresses a problem of sweep coverage by deploying a network of autonomous mobile robots. We propose a decentralized control algorithm for the robots to accomplish the sweep coverage. The sweep coverage is achieved by coordinating the robots to move along a given path that is unknown to the vehicles a priori. The motion coordination algorithm is developed based on simple consensus algorithms. The effectiveness of the algorithm is demonstrated via numerical simulations. The proposed algorithm would have applications to military and civilian operations.     

安全强化学习算法及其在CPS智能控制中的应用

信息物理系统(cyber-physicalsystem,CPS)的安全控制器设计是一个热门研究方向,现有基于形式化方法的安全控制器设计存在过度依赖模型、可扩展性差等问题.基于深度强化学习的智能控制可处理高维非线性复杂系统和不确定性系统,正成为非常有前景的CPS控制技术,但是缺乏对安全性的保障.针对强化学习控制在安全性方面的不足,围绕一个工业油泵控制系统典型案例,开展安全强化学习算法和智能控制应用研究.首先,形式化了工业油泵控制的安全强化学习问题,搭建了工业油泵仿真环境;随后,通过设计输出层结构和激活函数,构造了神经网络形式的油泵控制器,使得油泵开关时间的线性不等式约束得到满足;最后,为了更好地权衡安全性和最优性控制目标,基于增广拉格朗日乘子法设计实现了新型安全强化学习算法.在工业油泵案例上的对比实验表明,该算法生成的控制器在安全性和最优性上均超越了现有同类算法.在进一步评估中,所生成神经网络控制器以90%的概率通过了严格形式化验证;同时,与理论最优控制器相比实现了低至2%的最优目标值损失.所提方法有望推广至更多应用场景,实例研究的方案有望为安全智能控制和形式化验证领域其他学者提供借鉴.     

快速路匝道入口智能网联车协同合并控制研究

针对快速路匝道入口场景在高车流量的情况下容易发生交通拥堵的问题,提出了一种快速路匝道入口智能网联车(connected and automated vehicles, CAV)协同合并控制的解决方案,将问题解耦成多车顺序决策和车辆运动规划两部分。其中多车顺序决策对通行效率起到重要作用,因此设计了一种基于状态评价模型(state evaluation model, SEM)的多车顺序决策算法。该算法首先建立状态空间并初始化,考虑通行效率和车辆延迟设计状态评价函数,通过状态转移关系选择出最优状态,最终回溯得到最优通行顺序。根据车辆状态和到达冲突点时间,控制器实时推导各车辆纵向速度的解析解,实现车辆运动规划。仿真和实验结果表明,该方案在满足交通系统实时性要求的同时能有效提高快速路匝道入口的通行效率,缩短车辆延迟,降低燃油消耗。     

Distributed stochastic multi-vehicle routing in the Euclidean plane with no communications

This paper presents an algorithm for the multi-vehicle routing problem with no communications among the vehicles. The scenario consists in a convex Euclidean mission space, where targets are generated according to a Poisson distribution in time and to a generic continuous spatial distribution. The targets must be visited by the vehicles, which, therefore, must act in coordination. Even if no communications are required, the proposed routing strategy succeeds in effectively partitioning the mission space among the vehicles: at low target generation rates, the algorithm leads to the well-known centroidal Voronoi tessellation, whereas at high target generation rates, simulation results show that it has better performances with respect to a reference algorithm with no communications among vehicles.     

基于路径跟踪的拖车式移动机器人动态窗口法

针对目前局部路径规划算法只适用于单车体机器人的问题,提出了一种针对拖车式移动机器人的动态窗口法。首先,利用多车体结构的路径跟踪方程实现对拖车式移动机器人的运动控制;然后,利用评价函数同时对牵引车和拖车进行评价并根据权重相加;最后,针对拖车结构特性,添加了运动过程中牵引车与拖车的夹角约束,保证运动轨迹的稳定性。仿真实验表明:拖车式移动机器人的运动控制可满足收敛性,同时所提算法实现了拖车式移动机器人局部路径规划的任务,且在运动过程中夹角变化均未超出限制。该研究对拖车式移动机器人的自主导航有极大的参考价值。     

A novel coordinated path planning method using k-degree smoothing for multi-UAVs

Coordinated path planning for multiple unmanned aerial vehicles (multi-UAVs) is a highly significant problem encountered in their coordinated control. In the interests of completing mission securely and efficiently, the advanced multi-UAVs control technology requires a universal smoothing method as well as a precise coordination strategy. In this paper, we propose a novel multi-UAVs coordinated path planning method based on the k-degree smoothing, a more complex environment consists of multiple threat sources of which is constructed. By employing the Improved Ant Colony Optimization algorithm, a k-degree smoothing method is also presented aiming at obtaining a more flyable path. Additionally, the multi-UAVs coordination algorithm is induced by k-degree smoothing, allowing the UAVs to arrive at the destination simultaneously or in an acceptable time interval. Finally, simulations of the comparison between the Improved Ant Colony Optimization and classic algorithm, the detailed smoothing method, and the coordination are respectively conducted to validate that the proposed approach is feasible and effective in multi-UAVs coordinated path planning problems.