Wright Laboratory, at Tyndall AFB, Florida, has contracted the University of Florida to develop autonomous navigation systems for a variety of robotic vehicles, capable of performing tasks associated with the location and removal of bombs and mines. One of the tasks involves surveying closed target ranges for unexploded buried munitions. Accuracy in path following is critical to the task. There are hundreds of acres that currently require surveying. The sites are typically divided into regions, where each mission can take up to 4.5 hours. These sites are usually surveyed along parallel rows. By improving the accuracy of path following, the distance between the rows can be increased to nearly the detection with of the ground penetrating sensors, resulting in increased acreage surveyed per mission. This paper evaluates a high-level PID and a pure pursuit steering controller. The controllers were combined into a weighted solution so that the desirable features of each controller is preserved. This strategy was demonstrated in simulation and implemented on a Navigation Test Vehicle. For a test path of varying curvature, the average lateral control error was 2 cm at a vehicle sped of 1.34 m/s.

Outdoor navigation poses a challenge since uneven road surface, sunshine, clutter background will cause problems for odometry, laser, and vision sensors. To improve position accuracy, single strip retroreflective beacons have been used in the localization process. But to match observed beacons during the motion of the robot is a problem since all beacons are identical. Also strong reflective objects in outdoors may cause false reading.In this paper, we describe how we use the extended Kalman filter algorithm to integrate data scanned from the laser scanner rotating at 2Hz and readings from other sensors. The results obtained from outdoor navigation are presented.

This paper presents two different topologies for an IR sensor system for line tracking in a mobile robot. One topology is based on using analog data from the sensor system and the other one uses discrete information. Comparative analysis of the two systems is provided.

Parametric mapping (PM) lies midway between older and proven artificial landmark based guidance systems and yet to be realized vision based guidance systems. It is a simple yet effective natural landmark recognition system offering freedom from the need for enhancements to the environment. Development of PM systems can be inexpensive and rapid and they are starting to appear in commercial and industrial applications. Together with a description of the structural framework developed to generically describe robot mobility, this paper illustrates clearly the parts of any mobile robot navigation and guidance system and their interrelationships. Among other things, the importance of the richness of the reference map, and not necessarily the sensor map, is introduced, the benefits of dynamic path planners to alleviate the need for separate object avoidance, and the independence of the PM system to the type of sensor input is shown.

A scheme for extracting environment features from 1D range data and their interpretation is presented. Segmentation is done by deciding on a measure of model fidelity which is applied to adjacent groups of measurements. The extraction process is considered to include a subsequent matching step where segments which belong to the same landmark ar to be merged while keeping track of those which originate from distinct features. This is done by an agglomerative hierarchical clustering algorithm with a Mahalanobis distance matrix. The method is discussed with straight line segments which are found in a generalized least squares sense using polar coordinates including their first-order covariance estimates. As a consequence, extraction is no longer a real time problem on the level of single range readings, but must be treated on the level of whole scans. Experimental results with three commercially available laser scanners are presented. The implementation on a mobile robot which performs a map-based localization demonstrate the accuracy and applicability of the method under real time conditions. The collection of line segments and associated covariance matrices obtained from the extraction process contains more information about the scene than is required for map-based localization. In a subsequent reasoning step this information is made explicit. By successive abstraction and consequent propagation of uncertainties, a compact scene model is finally obtained in the form of a weighted symbolic description preserving topology information and reflecting the main characteristics of a local observation.

There has been a significant increase in the use of Autonomous Guided Vehicles in ports, mines and other primary industries. Many of these applications require vehicles which operate safely and efficiently in unstructured environments at speeds approaching those of human- controlled vehicles. Meeting these objectives is extremely difficult and arguably one of the most important requirements is an accurate and robust localization system. In this paper we describe the development of a prototype, Kalman filter-based localization system for a conventional road vehicle operating in an outdoor environment at speeds in excess of 15 ms^-1. Using sparsely placed beacons, vehicle position can be resolved to the order of a meter. Three main themes are addressed. The first is a quantitative methodology for sensor suite design. Sensors are classified according to their frequency responses and the suite is chosen to ensure a uniform response across the spectrum of vehicle maneuvers. The second theme develops accurate, high-order nonlinear models of vehicle motion which incorporate kinematics, dynamics and slip due to type deformation. Each model is useful within a certain operating regime. Outside of this regime the model can fail and the third theme avoids this problem through the use multiple models algorithms which synergistically fuse the properties of several models. Through addressing these themes we have developed a navigation system which as been shown to be accurate and robust to different types of road surface and occasional sensor data loss. The theories and principles developed in this paper are being used to develop a navigation system for commercial mining vehicles.

In this theoretical paper we address a robot motion planning problem with a special type of uncertainty. This problem arises, if the metric of the planning space is incompletely known. The multiplicative and stochastic nature of the optimization criterium leads to some new and interesting effects.

This work prescribes the procedures that are required to implement, on a conventional mobile robot, a sensor based motion planning algorithm based on the generalized Voronoi graph (GVG). The GVG is a roadmap of a static environment; recall that a roadmap is a 1D representation of an environment which the robot can use to plan a path between any two points in that environment. Once the robot has constructed the roadmap, it has in essence explored the environment. This work describes some issues in incrementally constructing the GVG with a mobile robot wit a ring of sonar sensors. Specifically, we consider some issues in specularity and dead-reckoning error reduction.

We propose to use an analog neural network controller implemented in hardware, independent of the active control system, for use in a satellite backup control mode. The controller uses coarse sun sensor inputs. The field-of-view of the sensors activate the neural controller, creating an analog dead band with respect to the direction of the sun on each axis. This network controls the orientation of the vehicle toward the sunlight to ensure adequate power for the system. The attitude of the spacecraft is stabilized with respect to the ambient magnetic field on orbit. This paper develops a model of the controller using real-time coarse sun sensor data and a dynamic model of a prototype system based on a satellite system. The simulation results and the feasibility of this control method for use in a satellite backup control mode are discussed.

The objective of path planing is to find a sequence of states that a system has to visit in order to attain the goal state. Because of their real-time efficiency, potential field methods present a powerful heuristic to guide this search. However, potential field approaches can not guarantee goal attainability. They are often referred to as 'local methods' and are used in conjunction with a global path planning method to ensure completeness of the path planning algorithm. The present work introduces a novel methodology for path planing which combines the real- time efficiency of potential field methods with goal-attainability characteristics of global methods. The algorithm of this work is: 1) free from local minima, ii) capable of considering arbitrary-shaped obstacles, iii) computationally less complex than previous search methods; and iv) able to handle obstacle avoidance and goal attainability at the same time. At the first step a new probabilistic scheme, based on absorbing Markov chains, is presented for global planning inside structured environments, such as office, etc. The potential field method is then reformulated for adaptive path planning among modeled and new obstacles.

Communications and communications protocols will play an important role in mobile robot systems able to address real world applications. A poorly integrated 'stack' of communications protocols, or protocols which are poorly matched to the functional and performance characteristics of the underlying physical communications links, can greatly reduce the effectiveness of an otherwise well implemented robotic or networked sensors system. The proliferation of Internet-like networks in military as well as civilian domains has motivated research to address some of the performance limitations TCP suffers when using RF and other media with long bandwidth-delay, dynamic connectivity, and error-prone links. Beyond these performance issues, however, TCP is poorly matched to the requirements of mobile robot and other quasi-autonomous systems: it is oriented to providing a continuous data stream, rather than discrete messages, and the canonical 'socket' interface conceals short losses of communications connectivity, but simply gives up and forces the application layer software to deal with longer losses. For the multipurpose security and surveillance mission platform project, a software applique is being developed that will run on top of user datagram protocol to provide a reliable message-based transport service. In addition, a session layer protocol is planned to support the effective transfer of control of multiple platforms among multiple stations.

Developing the technologies suitable of ra high level robotic application such as cleaning a floor has proved extremely difficult. Developing the robot mobility technology has been a stumbling block and developing and integrating the applications technology to the machine and the mobility technology has also been a difficult stage in this quest, but doing so in a cost effective and realistic manner suitable for the market place and to compete with humans and manually operated machines has been the most difficult of all. This paper describes one of these quests spanning a 14 year period and resulting in what is hoped will be the world's first commercially manufactured household robot vacuum cleaner.

Autonomous and semi-autonomous full-sized ground vehicles are becoming increasingly important, particularly in military applications. Here we describe the instrumentation of one such vehicle, a 4-wheel drive Hummer, for autonomous robotic operation. Actuators for steering, brake, and throttle have been implemented on a commercially available Hummer. Control is provided by on-board and remote computation. On-board computation includes a PC-based control computer coupled to feedback sensors for he steering wheel, brake, and forward speed; and a Unix workstation for high-level control. A radio link connects the on- board computers to an operator's remote workstation running the Georgia Tech MissionLab system. The paper describes the design and implementation of this integrated hardware/software system that translates a remote human operator's commands into directed motion of the vehicle telerobotic control of the Hummer has been demonstrated in outdoor experiments.

In this paper, behavior-based cognitive control for special shape mobile robots is studied thoroughly by using reinforcement learning idea. The study case considered is that a cross mobile robot with the shape of a Greek cross '+', which is equipped with a perception system for sensing environment, passes through a crack whose width is less than the width and length of robot body. This means that the robot can never pass through the narrow crack unless it knows how to take advantage of its self special shape intelligently. The objective of our work is to investigate how to automatically generate a trajectory for the mobile robot based on its own behavior. In the paper we first briefly introduce a behavior-based cognitive control system developed in Windows platforms, by which we can design and study various behavior-based cognitive control schemes for different shape mobile robots. Secondly, we discus the definitions of state and action space for the behavior-based cognitive control, and then investigate a learning algorithm using reinforcement learning idea. Thirdly, we propose a behavior-based cognitive control architecture. Finally, in order to prove the validity and efficiency of the developed behavior-based control strategy, we present some simulation results for the cross mobile robot.