Mobile robot navigation in a partially structured static environment, using neural predictive control

This paper presents a way of implementing a model-based predictive controller (MBPC) for mobile robot navigation when unexpected static obstacles are present in the robot environment. The method uses a nonlinear model of mobile robot dynamics, and thus allows an accurate prediction of the future trajectories. An ultrasonic ranging system has been used for obstacle detection. A multilayer perceptron is used to implement the MBPC, allowing real-time implementation and also eliminating the need for high-level data sensor processing. The perceptron has been trained in a supervised manner to reproduce the MBPC behaviour. Experimental results obtained when applying the neural-network controller to a TRC Labmate mobile robot are given in the paper.     

Kinematic path‐tracking of mobile robot using iterative learning control

This paper develops a kinematic path‐tracking algorithm for a nonholonomic mobile robot using an iterative learning control (ILC) technique. The proposed algorithm produces a robot velocity command, which is to be executed by the proper dynamic controller of the robot. The difference between the velocity command and the actual velocity acts as state disturbances in the kinematic model of the mobile robot. Given the kinematic model with state disturbances, we present an ILC‐based path‐tracking algorithm. An iterative learning rule with both predictive and current learning terms is used to overcome uncertainties and the disturbances in the system. It shows that the system states, outputs, and control inputs are guaranteed to converge to the desired trajectories with or without state disturbances, output disturbances, or initial state errors. Simulations and experiments using an actual mobile robot verify the feasibility and validity of the proposed learning algorithm. © 2005 Wiley Periodicals, Inc.     

Sensor planning for mobile robot localization using Bayesian network inference

We propose a new method of sensor planning for mobile robot localization using Bayesian network inference. Since we can model causal relations between situations of the robot's behavior and sensing events as nodes of a Bayesian network, we can use the inference of the network for dealing with uncertainty in sensor planning and thus derive appropriate sensing actions. In this system we employ a multi-layered-behavior architecture for navigation and localization. This architecture effectively combines mapping of local sensor information and the inference via a Bayesian network for sensor planning. The mobile robot recognizes the local sensor patterns for localization and navigation using a learned regression function. Since the environment may change during the navigation and the sensor capability has limitations in the real world, the mobile robot actively gathers sensor information to construct and reconstruct a Bayesian network, and then derives an appropriate sensing action which maximizes a utility function based on inference of the reconstructed network. The utility function takes into account belief of the localization and the sensing cost. We have conducted some simulation and real robot experiments to validate the sensor planning system.     

Virtual sensors for human concepts—Building detection by an outdoor mobile robot

In human–robot communication it is often important to relate robot sensor readings to concepts used by humans. We suggest the use of a virtual sensor (one or several physical sensors with a dedicated signal processing unit for the recognition of real world concepts) and a method with which the virtual sensor can learn from a set of generic features. The virtual sensor robustly establishes the link between sensor data and a particular human concept. In this work, we present a virtual sensor for building detection that uses vision and machine learning to classify the image content in a particular direction as representing buildings or non-buildings. The virtual sensor is trained on a diverse set of image data, using features extracted from grey level images. The features are based on edge orientation, the configurations of these edges, and on grey level clustering. To combine these features, the AdaBoost algorithm is applied. Our experiments with an outdoor mobile robot show that the method is able to separate buildings from nature with a high classification rate, and to extrapolate well to images collected under different conditions. Finally, the virtual sensor is applied on the mobile robot, combining its classifications of sub-images from a panoramic view with spatial information (in the form of location and orientation of the robot) in order to communicate the likely locations of buildings to a remote human operator.     

Visual Recognition of Types of Corridor Segments for Mobile Robots

This paper presents a visual recognition method to identify types of corridor segments such as T-junctions, L-junctions, and dead ends using vanishing point-based visual features and a two-layer recognition framework. This approach is useful for efficient robot navigation in the sense that a mobile robot is able to recognize the corridor segment type before reaching it, allowing the robot to make navigation decisions in advance. Furthermore, owing to our novel visual features made by using nonvertical vanishing points satisfying Manhattan world assumption, it is more probable for a mobile robot to recognize corridor segment types under partial occlusion by human. Experimental results have also been provided to demonstrate the validity of the proposed approach in real world environments.     

Neural network learning from demonstration and epipolar geometry for visual control of a nonholonomic mobile robot

The control of a robot system using camera information is a challenging task regarding unpredictable conditions, such as feature point mismatch and changing scene illumination. This paper presents a solution for the visual control of a nonholonomic mobile robot in demanding real world circumstances based on machine learning techniques. A novel intelligent approach for mobile robots using neural networks (NNs), learning from demonstration (LfD) framework, and epipolar geometry between two views is proposed and evaluated in a series of experiments. A direct mapping from the image space to the actuator command is conducted using two phases. In an offline phase, NN–LfD approach is employed in order to relate the feature position in the image plane with the angular velocity for lateral motion correction. An online phase refers to a switching vision based scheme between the epipole based linear velocity controller and NN–LfD based angular velocity controller, which selection depends on the feature distance from the pre-defined interest area in the image. In total, 18 architectures and 6 learning algorithms are tested in order to find optimal solution for robot control. The best training outcomes for each learning algorithms are then employed in real time so as to discover optimal NN configuration for robot orientation correction. Experiments conducted on a nonholonomic mobile robot in a structured indoor environment confirm an excellent performance with respect to the system robustness and positioning accuracy in the desired location.     

Behavior-based intelligent mobile robot using an immunized reinforcement adaptive learning mechanism

In this paper, a novel immunized reinforcement adaptive learning mechanism employing a behavior-based knowledge and the on-line adapting capabilities of the immune system is proposed and applied to an intelligent mobile robot. Rather than building a detailed mathematical model of immune systems, we try to explore principles in the immune system focusing on its self-organization, adaptive capability and immune memory. Two levels of the immune system, underlying the ‘micro’ level of cell interactions, and emergent ‘macro’ level of the behavior of the system are investigated.To evaluate the proposed immunized architecture, a ‘food foraging work’ simulation environment containing a mobile robot, foods, with/without obstacles is created to simulate the real world. The simulation results validate several significant characteristics of the immunized architecture: adaptability, learning, self-organizing, and stable ecological niche approaching.     

一种基于连接机制和时空经验的认知地图学习与导航方法

提出了一种连接主义方法, 利用移动机器人自身的时空经验, 在缺乏全局坐标信息和环境先验模型的情况下, 建立面向目标的认知地图. 在线形成的时序处理网络 (TSPN)可提供简洁的历史感知信息, 以神经元激活特性保存空间知识, 引导机器人运动. 结合TSPN和反应式行为模块的导航系统可实现动态的路标及方向检测、路径学习和实时导航功能. 仿真和实际实验验证了系统的有效性和适应性.     

Effective landmark placement for accurate and reliable mobile robot navigation

Being able to navigate accurately is one of the fundamental capabilities of a mobile robot to effectively execute a variety of tasks including docking, transportation, and manipulation. As real-world environments often contain changing or ambiguous areas, existing features can be insufficient for mobile robots to establish a robust navigation behavior. A popular approach to overcome this problem and to achieve accurate localization is to use artificial landmarks. In this paper, we consider the problem of optimally placing such artificial landmarks for mobile robots that repeatedly have to carry out certain navigation tasks. Our method aims at finding the minimum number of landmarks for which a bound on the maximum deviation of the robot from its desired trajectory can be guaranteed with high confidence. The proposed approach incrementally places landmarks utilizing linearized versions of the system dynamics of the robot, thus allowing for an efficient computation of the deviation guarantee. We evaluate our approach in extensive experiments carried out both in simulations and with real robots. The experiments demonstrate that our method outperforms other approaches and is suitable for long-term operation of mobile robots.     

基于动态系统的机器人模仿学习方法研究

针对当前机器人模仿学习过程中,运动模仿存在无法收敛到目标点以及泛化能力差的问题,引入一种基于动态系统（dynamical system,DS）的模仿学习方法。该方法通过高斯混合模型（gaussian mixture model,GMM）将示教运动数据建模为一非线性动态系统;将DS全局稳定的充分条件作为约束,以保证DS所生成的所有轨迹收敛到目标点;将动态系统模型的参数学习问题转化为求解一个约束优化问题,从而得到模型参数。以7bot机械臂为实验对象,进行仿真实验和机器人实验,实验结果表明:该方法学习的DS模型从不同起点生成的所有轨迹都收敛到目标点,轨迹平滑,泛化能力好。     

Incremental online sparsification for model learning in real-time robot control

For many applications such as compliant, accurate robot tracking control, dynamics models learned from data can help to achieve both compliant control performance as well as high tracking quality. Online learning of these dynamics models allows the robot controller to adapt itself to changes in the dynamics (e.g., due to time-variant nonlinearities or unforeseen loads). However, online learning in real-time applications - as required in control - cannot be realized by straightforward usage of off-the-shelf machine learning methods such as Gaussian process regression or support vector regression. In this paper, we propose a framework for online, incremental sparsification with a fixed budget designed for fast real-time model learning. The proposed approach employs a sparsification method based on an independence measure. In combination with an incremental learning approach such as incremental Gaussian process regression, we obtain a model approximation method which is applicable in real-time online learning. It exhibits competitive learning accuracy when compared with standard regression techniques. Implementation on a real Barrett WAM robot demonstrates the applicability of the approach in real-time online model learning for real world systems.     

Adaptive unified motion control of mobile manipulators

This paper presents a unified motion controller for mobile manipulators which not only solves the problems of point stabilization and trajectory tracking but also the path following problem. The control problem is solved based on the kinematic model of the robot. Then, a dynamic compensation is considered based on a dynamic model with inputs being the reference velocities to the mobile platform and the manipulator joints. An adaptive controller for on-line updating the robot dynamics is also proposed. Stability and robustness of the complete control system are proved through the Lyapunov method. The performance of the proposed controller is shown through real experiments.     

Simultaneous exploration and segmentation for search and rescue

We consider the problem of active victim segmentation during a search‐and‐rescue (SAR) exploration mission. The robot is equipped with a multimodal sensor suite consisting of a camera, lidar, and pan‐tilt thermal sensor. The robot enters an unknown scene, builds a 3D model incrementally, and the proposed method simultaneously (a) segments the victims from incomplete multimodal measurements and (b) controls the motion of the thermal camera. Both of these tasks are difficult due to the lack of natural training data and the limited number of real‐world trials. In particular, we overcome the absence of training data for the segmentation task by employing a manually designed generative model, which provides a semisynthetic training data set. The limited number of real‐world trials is tackled by self‐supervised initialization and optimization‐based guiding of the motion control learning. In addition to that, we provide a quantitative evaluation of the proposed method on several real testing scenarios using the real SAR robot. Finally, we also provide a data set which will allow for further development of algorithms on the real data.     

Sensor-based fuzzy navigation of an autonomous mobile robot in an indoor environment

The work presented in this paper deals with the problem of navigating a mobile robot either in an unknown indoor environment or in a partially known one. A navigation method based on the combination of elementary behaviors has been developed for an unknown environment. Most of these behaviors are achieved by means of fuzzy inference systems. The proposed navigator combines two types of obstacle avoidance behaviors, one for the convex obstacles and one for the concave ones. In the case of a partially known environment, a hybrid method is used to exploit the advantages of global and local navigation strategies. The coordination of these strategies is based on a fuzzy inference system that involves an on-line comparison between the real scene and a memorized one. Both methods have been implemented on the miniature mobile robot Khepera^® which is equipped with rough sensors. The good results obtained illustrate the robustness of a fuzzy logic approach with regard to sensor imperfections.     

Deliberative On-Line Local Path Planning for Autonomous Mobile Robots

This paper describes a method for local path planning for mobile robots that combines reactive obstacle avoidance with on-line local path planning. Our approach is different to other model-based navigation approaches since it integrates both global and local planning processes in the same architecture while other methods only combine global path planning with a reactive method to avoid non-modelled obstacles. Our local planning is only triggered when an unexpected obstacle is found and reactive navigation is not able to regain the initial path. A new trajectory is then calculated on-line using only proximity sensor information. This trajectory can be improved during the available time using an anytime algorithm. The proposed method complements the reactive behaviour and allows the robot to navigate safely in a partially known environment during a long time period without human intervention.     

Sensor Fault Tolerance Method by Using a Bayesian Network for Robot Behavior

This paper presents FTBN, a new framework that performs learning autonomous mobile robot behavior and fault tolerance simultaneously. For learning behavior in the presence of a robot sensor fault this framework uses a Bayesian network. In the proposed framework, sensor data are used to detect a faulty sensor. Fault isolation is accomplished by changing the Bayesian network structure using interpreted evidence from robot sensors. Experiments including both simulation and a real robot are performed for door-crossing behavior using prior knowledge and sensor data at several maps. This paper explains the learning behavior, optimal tracking, exprimental setup and structure of the proposed framework. The robot uses laser and sonar sensors for door-crossing behavior, such that each sensor can be corrupted during the behavior. Experimental results show FTBN leads to robust behavior in the presence of a sensor fault as well as performing better compared to the conventional Bayesian method.     

Smart radiation sensor management

We developed two radiation mapping algorithms that can handle different situations based on prior information of the search area. The algorithms were developed in the framework of model-driven measurement, where a world model was used to drive measurement collection, and measurements were used to update the world model.We developed and experimentally tested a robotic implementation of two Bayesian-based radiation mapping strategies in two dimensions, using a commercially available desktop mobile robot fitted with a CsI radiation sensor. Our approach for implementing the Bayesian radiation mapping algorithms was to drive the robot over each segment of the search area, in real time, according to the radiation counts collected by the sensor. Future research directions include extensions to three-dimensional mapping; exploring and characterizing the tradeoffs between time efficiency, map confidence level, and utilization of prior knowledge information; as well as the implementation of Bayesian statistics for the online update of the world model.     

Segment-Based Map Building Using Enhanced Adaptive Fuzzy Clustering Algorithm for Mobile Robot Applications

In this paper, we present a technique for on-line segment-based map building in an unknown indoor environment from sonar sensor observations. The world model is represented with two-dimensional line segments. The information obtained by the ultrasonic sensors is updated instantaneously while the mobile robot is moving through the workspace. An Enhanced Adaptive Fuzzy Clustering Algorithm (EAFC) along with Noise Clustering (NC) is proposed to extract and classify the line segments in order to construct a complete map for an unknown environment. Furthermore, to alleviate the problem of extensive computation associated with the process of map building, the workplace of the mobile robot is divided into square cells. A compatible line segment merging technique is then suggested to combine the similar segments after the extraction of the line segment by EAFC along with NC algorithm. The performance of the algorithm is demonstrated by experimental results on a Pioneer II mobile robot.     

Simultaneous localization and map building for mobile robotnavigation

Inexpensive ultrasonic sensors, incremental encoders, and grid-based probabilistic modeling are used for improved robot navigation in indoor environments. For model-building, range data from ultrasonic sensors are constantly sampled and a map is built and updated immediately while the robot is travelling through the workspace. The local world model is based on the concept of an occupancy grid. The world model extracted from the range data is based on the geometric primitive of line segments. For the extraction of these features, methods such as the Hough transform and clustering are utilized. The perceived local world model along with dead-reckoning and ultrasonic sensor data are combined using an extended Kalman filter in a localization scheme to estimate the current position and orientation of the mobile robot, which is subsequently fed to the map-building algorithm. Implementation issues and experimental results with the Nomad 150 mobile robot in a real-world indoor environment (office space) are presented     

Life-long optimization of the symbolic model of indoor environments for a mobile robot.

The use of a symbolic model of the spatial environment becomes crucial for a mobile robot that is intended to operate optimally and intelligently in indoor scenarios. Constructing such a model involves important problems that are not solved completely at present. One is called anchoring, which implies to maintain a correct dynamic correspondence between the real world and the symbols in the model. The other problem is adaptation: among the numerous possible models that could be constructed for representing a given environment, optimization involves the selection of one that improves as much as possible the operations of the robot. To cope with both problems, in this paper, we propose a framework that allows an indoor mobile robot to learn automatically a symbolic model of its environment and to optimize it over time with respect to changes in both the environment and the robot operational needs through an evolutionary algorithm. For coping efficiently with the large amounts of information that the real world provides, we use abstraction, which also helps in improving task planning. Our experiments demonstrate that the proposed framework is suitable for providing an indoor mobile robot with a good symbolic model and adaptation capabilities.