Maximizing Reward in a Non-Stationary Mobile Robot Environment

The ability of a robot to improve its performance on a task can be critical, especially in poorly known and non-stationary environments where the best action or strategy is dependent upon the current state of the environment. In such systems, a good estimate of the current state of the environment is key to establishing high performance, however quantified. In this paper, we present an approach to state estimation in poorly known and non-stationary mobile robot environments, focusing on its application to a mine collection scenario, where performance is quantified using reward maximization. The approach is based on the use of augmented Markov models (AMMs), a sub-class of semi-Markov processes. We have developed an algorithm for incrementally constructing arbitrary-order AMMs on-line. It is used to capture the interaction dynamics between a robot and its environment in terms of behavior sequences executed during the performance of a task. For the purposes of reward maximization in a non-stationary environment, multiple AMMs monitor events at different timescales and provide statistics used to select the AMM likely to have a good estimate of the environmental state. AMMs with redundant or outdated information are discarded, while attempting to maintain sufficient data to reduce conformation to noise. This approach has been successfully implemented on a mobile robot performing a mine collection task. In the context of this task, we first present experimental results validating our reward maximization performance criterion. We then incorporate our algorithm for state estimation using multiple AMMs, allowing the robot to select appropriate actions based on the estimated state of the environment. The approach is tested first with a physical robot, in a non-stationary environment with an abrupt change, then with a simulation, in a gradually shifting environment.     

3D Collision-Free Motion Based on Collision Index

A collision-free motion planning method for mobile robots moving in 3-dimensional workspace is proposed in this article. To simplify the mathematical representation and reduce the computation complexity for collision detection, objects in the workspace are modeled as ellipsoids. By means of applying a series of coordinate and scaling transformations between the robot and the obstacles in the workspace, intersection check is reduced to test whether the point representing the robot falls outside or inside the transformed ellipsoids representing the obstacles. Therefore, the requirement of the computation time for collision detection is reduced drastically in comparison with the computational geometry method, which computes a distance function of the robot segments and the obstacles. As a measurement of the possible occurrence of collision, the collision index, which is defined by projecting conceptually an ellipsoid onto a 3-dimensional Gaussian distribution contour, plays a significant role in planning the collision-free path. The method based on reinforcement learning search using the defined collision index for collision-free motion is proposed. A simulation example is given in this article to demonstrate the efficiency of the proposed method. The result shows that the mobile robot can pass through the blocking obstacles and reach the desired final position successfully after several trials.     

Optimal Robot Localization in Trees

The problem of localization, that is, of a robot finding its position on a map, is an important task for autonomous mobile robots. It has applications in numerous areas of robotics ranging from aerial photography to autonomous vehicle exploration. In this paper we present a new strategy LPS (Localize-by-Placement-Separation) for a robot to find its position on a map, where the map is represented as a geometric tree of bounded degree. Our strategy exploits to a high degree the self-similarities that may occur in the environment. We use the framework of competitive analysis to analyze the performance of our strategy. In particular, we show that the distance traveled by the robot is at most O( ) times longer than the shortest possible route to localize the robot, where n is the number of vertices of the tree. This is a significant improvement over the best known previous bound of O(n^2/3). Moreover, since there is a lower bound of Ω( ), our strategy is optimal up to a constant factor. Using the same approach we can also show that the problem of searching for a target in a geometric tree, where the robot is given a map of the tree and the location of the target but does not know its own position, can be solved by a strategy with a competitive ratio of O( ), which is again optimal up to a constant factor.     

Mobile Robot Self-Localization without Explicit Landmarks

Localization is the process of determining the robot's location within its environment. More precisely, it is a procedure which takes as input a geometric map, a current estimate of the robot's pose, and sensor readings, and produces as output an improved estimate of the robot's current pose (position and orientation). We describe a combinatorially precise algorithm which performs mobile robot localization using a geometric model of the world and a point-and-shoot ranging device. We also describe a rasterized version of this algorithm which we have implemented on a real mobile robot equipped with a laser rangefinder we designed. Both versions of the algorithm allow for uncertainty in the data returned by the range sensor. We also present experimental results for the rasterized algorithm, obtained using our mobile robots at Cornell. Received November 15, 1996; revised January 13, 1998.     

Motion Planning Strategy for Finding an Object with a Mobile Manipulator in Three-Dimensional Environments

In this paper, we investigate the problem of minimizing the average time required to find an object in a known three-dimensional environment. We consider a 7-d.o.f. mobile manipulator with an 'eye-in-hand' sensor. In particular, we address the problem of searching for an object whose unknown location is characterized by a known probability density function. We present a discrete formulation, in which we use a visibility-based decomposition of the environment. We introduce a sample-based convex cover to estimate the size and shape of visibility regions in three dimensions. The resulting convex regions are exploited to generate trajectories that make a compromise between moving the manipulator base and moving the robotic arm. We also propose a practical method to approximate the visibility region in three dimensions of a sensor limited in both range and field of view. The quality and success of the generated paths depend significantly on the sensing robot capabilities. In this paper, we generate searching plans for a mobile manipulator equipped with a sensor limited in both field of view and range. We have implemented the algorithm and present simulation results.     

Real-time Velocity Alteration Strategy for Collision-free Trajectory Planning of Two Articulated Robot Manipulators

Two articulated robots working in a shared workspace can be programmed by planning the tip trajectory of each robot independently. To account for collision avoidance between links, a real-time velocity alteration strategy based on fast and accurate collision detection is proposed in this paper to determine the step of next motion of slave (low priority) robot for collision-free trajectory planning of two robots with priorities. The effectiveness of the method depends largely on a newly developed method of accurate estimate of distance between links. By using the enclosing and enclosed ellipsoids representations of polyhedral models of links of robots, the minimum distance estimate and collision detection between the links can be performed more efficiently and accurately. The proposed strategy is implemented in an environment where the geometric paths of robots are pre-planned and the preprogrammed velocities are piecewise constant but adjustable. Under the control of the proposed strategy, the master robot always moves at a constant speed. The slave robot moves at the selected velocity, selected by a tradeoff between collision trend index and velocity reduction in one collision checking time, to keep moving as far as possible and as fast as possible while avoid possible collisions along the path. The collision trend index is a fusion of distance and relative velocity between links of two robots to reflect the possibility of collision at present and in the future. Graphic simulations of two PUMA560 robot arms working in common workspace but with independent goals are conducted. Simulations demonstrate the collision avoidance capability of the proposed approach as compared to the approach based on bounding volumes. It shows that advantage of our approach is less number of speed alterations required to react to potential collisions.     

Motion planning in uncertain environments with vision-like sensors

In this work we present a methodology for intelligent path planning in an uncertain environment using vision-like sensors, i.e., sensors that allow the sensing of the environment non-locally. Examples would include a mobile robot exploring an unknown terrain or a micro-UAV navigating in a cluttered urban environment. We show that the problem of path planning in an uncertain environment, under certain assumptions, can be posed as the adaptive optimal control of an uncertain Markov decision process, characterized by a known, control-dependent system, and an unknown, control-independent environment. The strategy for path planning then reduces to computing the control policy based on the current estimate of the environment, also known as the “certainty-equivalence” principle in the adaptive control literature. Our methodology allows the inclusion of vision-like sensors into the problem formulation, which, as empirical evidence suggests, accelerates the convergence of the planning algorithms. Further we show that the path planning and estimation problems, as formulated in this paper, possess special structure which can be exploited to significantly reduce the computational burden of the associated algorithms. We apply this methodology to the problem of path planning of a mobile rover in a completely unknown terrain.     

Concurrent localization of multiple robots

A concurrent localization method for multiple robots using ultrasonic beacons is proposed. This method provides a high-accuracy solution using only low-price sensors. To measure the distance of a mobile robot from a beacon at a known position, the mobile robot alerts one beacon to send out an ultrasonic signal to measure the traveling time from the beacon to the mobile robot. When multiple robots requiring localization are moving in the same block, it is necessary to have a schedule to choose the measuring sequence in order to overcome constant ultrasonic signal interference among robots. However, the increased time delay needed to estimate the positions of multiple robots degrades the localization accuracy. To solve this problem, we propose an efficient localization algorithm for multiple robots, where the robots are in groups of one master robot and several slave robots. In this method, when a master robot calls a beacon, all the group robots simultaneously receive an identical ultrasonic signal to estimate their positions. The effectiveness of the proposed algorithm has been verified through experiments.     

Toward efficient path-planning for articulated robots

Real-time issues are becoming more and more important in robot programming. When a 6-dof manipulator is used, planning obstacle-avoiding paths is a time-consuming activity, usually done in simulation. We present the geometric models and the reasoning techniques we have implemented while realizing a gross motion planner for a manipulator with six revolute joints. First, construction of a problem-oriented representation of the robot working space is explained. Then, the actual trajectory research carried out in our C-space representation is described. The whole C-space is not calculated; instead, a sequential strategy is used to determine the C-space only for the first two links. Our approximation of the obstacles, which occupy fixed and known positions, greatly speeds the computation, allowing us to reduce the problem to planar geometric reasoning. The work is not limited to theoretical studies or simulations; experiments have been run very thoroughly, with various tests, on a PUMA robot to assess the real efficiency and usability of our software. The method applies to robots in a fixed and known environment. © 3995 John Wiley & Sons, Inc.     

可移动机器人在中心对称环境中的自定位算法

可移动机器人的自定位问题是智能机器人研究中的重要课题，它包含许多传感器技术和定位算法，马尔可夫定位算法的优点是可以使机器人在全局不确定的情况下估计它的位置。这种方法采用概率分布描述机器人的位置信度，机器人通过在运动过程中所获得的传感器数据和运动记录来更新信度分布，然后采用最高信度值来估计它所在的位置。对于只有距离测量传感器的机器人在中心对称环境中仅仅采用马尔可夫自定位法还是无法确定其位置，为了解决中心对称的环境中所存在的问题，建议在机器人上装上陀螺仪或指南针，定义一个角度高斯分布函数，并利用这个函数建立新的机器人感知模型来扩展马尔可夫定位算法，通过仿真程序对多种对称情况进行实验，验证了这一新算法的可行性，这个扩展马尔可夫自定位算法不仅可使机器人在中心对称环境中很快地确定自己的位置，而且可以加快非对称环境中信度分布收敛到真实位置的速度。     

Exploring Unknown Environments with Obstacles

Abstract. We study exploration problems where a robot has to construct a complete map of an unknown environment using a path that is as short as possible. In the first problem setting we consider, a robot has to explore n rectangles. We show that no deterministic or randomized online algorithm can be better than Ω(\sqrt n ) -competitive, solving an open problem by Deng et al. [7]. We also generalize this bound to the problem of exploring three-dimensional rectilinear polyhedra without obstacles. In the second problem setting we study, a robot has to explore a grid graph with obstacles in a piecemeal fashion. The piecemeal constraint was defined by Betke et al. [4] and implies that the robot has to return to a start node every so often. Betke et al. gave an efficient algorithm for exploring grids with rectangular obstacles. We present an efficient strategy for piecemeal exploration of grids with arbitrary rectilinear obstacles.     

Estimating graph distance and centrality on shared nothing architectures

We present a parallel toolkit for pairwise distance computation in massive networks. Computing the exact shortest paths between a large number of vertices is a costly operation, and serial algorithms are not practical for billion‐scale graphs. We first describe an efficient parallel method to solve the single source shortest path problem on commodity hardware with no shared memory. Using it as a building block, we introduce a new parallel algorithm to estimate the shortest paths between arbitrary pairs of vertices. Our method exploits data locality, produces highly accurate results, and allows batch computation of shortest paths with 7% average error in graphs that contain billions of edges. The proposed algorithm is up to two orders of magnitude faster than previously suggested algorithms and does not require large amounts of memory or expensive high‐end servers. We further leverage this method to estimate the closeness and betweenness centrality metrics, which involve systems challenges dealing with indexing, joining, and comparing large datasets efficiently. In one experiment, we mined a real‐world Web graph with 700 million nodes and 12 billion edges to identify the most central vertices and calculated more than 63 billion shortest paths in 6 h on a 20‐node commodity cluster. Copyright © 2014 John Wiley & Sons, Ltd.     

Piecemeal Graph Exploration by a Mobile Robot

We study how a mobile robot can learn an unknown environment in a piecemeal manner. The robot's goal is to learn a complete map of its environment, while satisfying the constraint that it must return every so often to its starting position (for refueling, say). The environment is modeled as an arbitrary, undirected graph, which is initially unknown to the robot. We assume that the robot can distinguish vertices and edges that it has already explored. We present a surprisingly efficient algorithm for piecemeal learning an unknown undirected graph G=(V, E) in which the robot explores every vertex and edge in the graph by traversing at most O(E+V^1+o(1)) edges. This nearly linear algorithm improves on the best previous algorithm, in which the robot traverses at most O(E+V²) edges. We also give an application of piecemeal learning to the problem of searching a graph for a “treasure.”     

Development of distance recognition using an ocellus camera for an autonomous personal robot

We are attempting to develop an autonomous personal robot that has the ability to perform practical tasks in a human living environment by using information derived from sensors. When a robot operates in a human environment, the issue of safety must be considered in regard to its autonomous movement. Thus, robots absolutely require systems that can recognize the external world and perform correct driving control. We have thus developed a navigation system for an autonomous robot. The system requires only image data captured by an ocellus CCD camera. In this system, we allow the robot to search for obstacles present on the floor. Then, the robot obtains distance recognition necessary for evasion of the object, including data of the obstacle’s width, height, and depth by calculating the angles of images taken by the CCD camera. We applied the system to a robot in an indoor environment and evaluated its performance, and we consider the resulting problems in the discussion of our experimental results. This work was presented in part at the 13th International Symposium on Artificial Life and Robotics, Oita, Japan, January 31–February 2, 2008     

Reliability-based camera handoff for cooperative tracking with multiple pan-tilt cameras

We present a new handoff method for multiple pan-tilt cameras for mobile robot tracking in an indoor environment. Camera handoff is an important step to consistently maintain the visibility of a mobile robot with maximized object tracking accuracy. First, we propose a method to estimate the position of a mobile robot using single pan-tilt camera. Then, the concept of position reliability is defined to quantitatively evaluate the accuracy of position estimation and tracking ability of individual pan-tilt cameras. Position reliability is used to decide when to trigger handoff and who to response handoff in the proposed handoff algorithm. Experimental results demonstrate that four pan-tilt cameras can systematically track a mobile robot in an indoor environment using the proposed method.     

Precise Hausdorff distance computation for planar freeform curves using biarcs and depth buffer

We present a real-time algorithm for computing the precise Hausdorff Distance (HD) between two planar freeform curves. The algorithm is based on an effective technique that approximates each curve with a sequence of G ¹ biarcs within an arbitrary error bound. The distance map for the union of arcs is then given as the lower envelope of trimmed truncated circular cones, which can be rendered efficiently to the graphics hardware depth buffer. By sampling the distance map along the other curve, we can estimate a lower bound for the HD and eliminate many redundant curve segments using the lower bound. For the remaining curve segments, we read the distance map and detect the pixel(s) with the maximum distance. Checking a small neighborhood of the maximum-distance pixel, we can reduce the computation to considerably smaller subproblems, where we employ a multivariate equation solver for an accurate solution to the original problem. We demonstrate the effectiveness of the proposed approach using several experimental results.     

Generalized measuring-worm algorithm: high-accuracy mapping and movement via cooperating swarm robots

Recently, many extensive studies have been conducted on robot control via self-positioning estimation techniques. In the simultaneous localization and mapping (SLAM) method, which is one approach to self-positioning estimation, robots generally use both autonomous position information from internal sensors and observed information on external landmarks. SLAM can yield higher accuracy positioning estimations depending on the number of landmarks; however, this technique involves a degree of uncertainty and has a high computational cost, because it utilizes image processing to detect and recognize landmarks. To overcome this problem, we propose a state-of-the-art method called a generalized measuring-worm (GMW) algorithm for map creation and position estimation, which uses multiple cooperating robots that serve as moving landmarks for each other. This approach allows problems of uncertainty and computational cost to be overcome, because a robot must find only a simple two-dimensional marker rather than feature-point landmarks. In the GMW method, the robots are given a two-dimensional marker of known shape and size and use a front-positioned camera to determine the marker distance and direction. The robots use this information to estimate each other’s positions and to calibrate their movement. To evaluate the proposed method experimentally, we fabricated two real robots and observed their behavior in an indoor environment. The experimental results revealed that the distance measurement and control error could be reduced to less than 3 %.     

Appearance-based mapping using minimalistic sensor models

This paper addresses the problem of localization and map construction by a mobile robot in an indoor environment. Instead of trying to build high-fidelity geometric maps, we focus on constructing topological maps as they are less sensitive to poor odometry estimates and position errors. We propose a modification to the standard SLAM algorithm in which the assumption that the robots can obtain metric distance/bearing information to landmarks is relaxed. Instead, the robot registers a distinctive sensor “signature”, based on its current location, which is used to match robot positions. In our formulation of this non-linear estimation problem, we infer implicit position measurements from an image recognition algorithm. We propose a method for incrementally building topological maps for a robot which uses a panoramic camera to obtain images at various locations along its path and uses the features it tracks in the images to update the topological map. The method is very general and does not require the environment to have uniquely distinctive features. Two algorithms are implemented to address this problem. The Iterated form of the Extended Kalman Filter (IEKF) and a batch-processed linearized ML estimator are compared under various odometric noise models.

Distributed information-based guidance of multiple mobile sensors for urban target search

We present a framework for distributed mobile sensor guidance to locate and track a target inside an urban environment. Our approach leverages the communications between robots when a link is available, but it also allows them to act independently. Each robot actively seeks the target using information maximization. The robots are assumed to be capable of communicating with their peers within some distance radius, and the sensor payload of each robot is a camera modeled to have target detection errors of types I and II. Our contributions include an optimal information fusion algorithm for discrete distributions which allows each agent to combine its local information with that of its neighbors, and a path planner that uses the fused estimate and a recent coverage result for information maximization to guide the agents. We include simulations and laboratory experiments involving multiple robots searching for a moving target within model cities of different sizes.     

Bridging the gap between feature- and grid-based SLAM

One important design decision for the development of autonomously navigating mobile robots is the choice of the representation of the environment. This includes the question of which type of features should be used, or whether a dense representation such as occupancy grid maps is more appropriate. In this paper, we present an approach which performs SLAM using multiple representations of the environment simultaneously. It uses reinforcement to learn when to switch to an alternative representation method depending on the current observation. This allows the robot to update its pose and map estimate based on the representation that models the surrounding of the robot in the best way. The approach has been implemented on a real robot and evaluated in scenarios, in which a robot has to navigate in- and outdoors and therefore switches between a landmark-based representation and a dense grid map. In practical experiments, we demonstrate that our approach allows a robot to robustly map environments which cannot be adequately modeled by either of the individual representations.