"Consider two people having a face-to-face conversation. They sometimes listen, sometimes talk, and sometimes interrupt each other. They use facial expressions to signal that they are confused. They point at objects. They jump from topic to topic opportunistically. When another acquaintance walks by, they nod and say hello. All the while they have other concerns on their mind, such as not missing the meeting that starts in 10 minutes. Like many other humans behaviors, these are not easy to replicate in artificial agents. In this work we look into the design requirements of an embodied agent that can participate in such natural conversations in a mixed-initiative, multi-modal setting. Such an agent needs to understand participating in a conversation is not merely a matter of sending a message and then waiting to receive a response -- both partners are simultaneously active at all times. This agent should be able to deal with different, sometimes conflicting goals, and be always ready to address events that may interrupt the current topic of conversation. To address those requirements, we have created a modular architecture that includes distributed functional units that compete with each other to gain control over available resources. Each of these units, called a schema, has its own sense- think-act cycle. In the field of robotics, this design is often referred to as "behavior-based" or "schema-based." The major contribution of this work is merging behavior-based robotics with plan- based human-computer interaction."

Abstract This thesis presents a novel control architecture, Samba, for reactive task execution. Reactive task execution implies goal-oriented and reactive properties from a robot and the ability to execute several tasks at the same time, also in a dynamic environment. These requirements are fullfilled in Samba by the rrepresentation of goals, intermediate results, and robots actions. The key idea in Samba is to produce continously reactions for all the important objects in the environment. These reactions are represented as action maps, which are a novel representation for robot actions. An action map specifies for each possible action how preferable the action is from the perspective of the producer of the map. the preferences are shown by assigning a weight to each action. Tasks are executed by modifying and combining action maps. The tasks can be either reasoned by a higher layer or triggered by sensor data. Action maps, and the methods for modifying and combining them, enable executing tasks inparallel and considering the dynamics of the environment. further, as the action maps are produced continously from sensor data, the robot actions are based on the current state of the environment. Markers describe goals and intermediate results. They facilitate managing the complexity of the system. Markers describing intermediate results decompose the system vertically, into producers and consumers of data. Markers describing goals decompose the control system horizontally, into a Samba layer and a higher layer of reasoning tasks. Tasks flow via markers from the higher layer to the Samba layer. Markers are tested on a real robot equipment with stereo gaze platform. Further, the samba architecture is applied to playing soccer. Experiments were carried out in the 1997 and 1998 RoboCup competitions. These experiments show that the Samba architecture is a potential alternative for controlling a mobile robot in a dynamic environment.

Fast spreading unknown viruses have caused major damage on computer systems upon their initial release. Current detection methods have lacked capabilities to detect unknown virus quickly enough to avoid mass spreading and damage. This dissertation has presented a behavior based approach to detecting known and unknown viruses based on their attempt to replicate. Replication is the qualifying fundamental characteristic of a virus and is consistently present in all viruses making this approach applicable to viruses belonging to many classes and executing under several conditions. A form of replication called self-reference replication, (SR-replication), has been formalized as one main type of replication which specifically replicates by modifying or creating other files on a system to include the virus itself. This replication type was used to detect viruses attempting replication by referencing themselves which is a necessary step to successfully replicate files. The approach does not require a priori knowledge about known viruses. Detection was accomplished at runtime by monitoring currently executing processes attempting to replicate. Two implementation prototypes of the detection approach called SRRAT were created and tested on the Microsoft Windows operating systems focusing on the tracking of user mode Win32 API system calls and Kernel mode system services. The research results showed SR-replication capable of distinguishing between file infecting viruses and benign processes with little or no false positives and false negatives.

@font-face { font-family: "TimesNewRoman"; }p.MsoNormal, li.MsoNormal, div.MsoNormal { margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: "Times New Roman"; }div.Section1 { page: Section1; } Teleoperated task execution for hazardous environments is slow and requires highly skilled operators. Attempts to implement telerobotic assists to improve efficiency have been demonstrated in constrained laboratory environments but are not being used in the field because they are not appropriate for use on actual remote systems operating in complex unstructured environments using typical operators. This work describes a methodology for combining select concepts from behavior-based systems with telerobotic tool control in a way that is compatible with existing manipulator architectures used by remote systems typical to operations in hazardous environment. The purpose of the approach is to minimize the task instance modeling in favor of a priori task type models while using sensor information to register the task type model to the task instance. The concept was demonstrated for two tools useful to decontamination & dismantlement type operations—a reciprocating saw and a powered socket tool. The experimental results demonstrated that the approach works to facilitate traded control telerobotic tooling execution by enabling difficult tasks and by limiting tool damage. The role of the tools and tasks as drivers to the telerobotic implementation was better understood in the need for thorough task decomposition and the discovery and examination of the tool process signature. The contributions of this work include: (1) the exploration and evaluation of select features of behavior-based robotics to create a new methodology for integrating telerobotic tool control with positional teleoperation in the execution of complex tool-centric remote tasks, (2) the simplification of task decomposition and the implementation of sensor-based tool control in such a way that eliminates the need for the creation of a task instance model for telerobotic task execution, and (3) the discovery, demonstrated use, and documentation of characteristic tool process signatures that have general value in the investigation of other tool control, tool maintenance, and tool development strategies above and beyond the benefit sustained for the methodology described in this work.

International Telemetering Conference Proceedings / October 18-21, 2004 / Town & Country Resort, San Diego, California / In this paper we propose an open multi-robot architecture that dramatically reduces the time to deployment and increases the utility value to the mainstream non-technical user. We describe a multi-robot behavior-based coordination architecture and argue its suitability in the context of general-purpose robot teams operating in dynamic and unpredictable environments. We then formalize and describe a command fusion module for the coordination of high-level behaviors of the system. The command fusion module is interfaced to our middle-ware/compiler that generates behavior selection tips from a user specified abstract description of a scenario. Finally, we utilize an example search and rescue scenario to illustrate the overall process and give preliminary results of the experiments performed on actual robots.

In this report, I discuss the use of vision to support concrete, everyday activity. I will argue that a variety of interesting tasks can be solved using simple and inexpensive vision systems. I will provide a number of working examples in the form of a state-of-the-art mobile robot, Polly, which uses vision to give primitive tours of the seventh floor of the MIT AI Laboratory. By current standards, the robot has a broad behavioral repertoire and is both simple and inexpensive (the complete robot was built for less than $20,000 using commercial board-level components). The approach I will use will be to treat the structure of the agent's activity---its task and environment---as positive resources for the vision system designer. By performing a careful analysis of task and environment, the designer can determine a broad space of mechanisms which can perform the desired activity. My principal thesis is that for a broad range of activities, the space of applicable mechanisms will be broad enough to include a number mechanisms which are simple and economical. The simplest mechanisms that solve a given problem will typically be quite specialized to that problem. One thus worries that building simple vision systems will be require a great deal of {it ad-hoc} engineering that cannot be transferred to other problems. My second thesis is that specialized systems can be analyzed and understood in a principled manner, one that allows general lessons to be extracted from specialized systems. I will present a general approach to analyzing specialization through the use of transformations that provably improve performance. By demonstrating a sequence of transformations that derive a specialized system from a more general one, we can summarize the specialization of the former in a compact form that makes explicit the additional assumptions that it makes about its environment. The summary can be used to predict the performance of the system in novel environments. Individual transformations can be recycled in the design of future systems.

This thesis presents methods for implementing robust hexpod locomotion on an autonomous robot with many sensors and actuators. The controller is based on the Subsumption Architecture and is fully distributed over approximately 1500 simple, concurrent processes. The robot, Hannibal, weighs approximately 6 pounds and is equipped with over 100 physical sensors, 19 degrees of freedom, and 8 on board computers. We investigate the following topics in depth: distributed control of a complex robot, insect-inspired locomotion control for gait generation and rough terrain mobility, and fault tolerance. The controller was implemented, debugged, and tested on Hannibal. Through a series of experiments, we examined Hannibal's gait generation, rough terrain locomotion, and fault tolerance performance. These results demonstrate that Hannibal exhibits robust, flexible, real-time locomotion over a variety of terrain and tolerates a multitude of hardware failures.

We introduce basic behaviors as primitives for control and learning in situated, embodied agents interacting in complex domains. We propose methods for selecting, formally specifying, algorithmically implementing, empirically evaluating, and combining behaviors from a basic set. We also introduce a general methodology for automatically constructing higher--level behaviors by learning to select from this set. Based on a formulation of reinforcement learning using conditions, behaviors, and shaped reinforcement, out approach makes behavior selection learnable in noisy, uncertain environments with stochastic dynamics. All described ideas are validated with groups of up to 20 mobile robots performing safe--wandering, following, aggregation, dispersion, homing, flocking, foraging, and learning to forage.

The objective of the master’s thesis work is to investigate how an autonomous underwater vehicle (AUV) should act in an underwater tunnel environment. The thesis proposes sensors, control strategies, mission statement, among others, required for tunnel assignments. A behavior-based control (BBC) strategy has been developed to control the AUV. The BBC is used in the middle level of the vehicle control, i.e. the reactive control system which describes how the AUV navigates through a tunnel, while other events are considered. The control strategy has also been separated into two parts, and these are: controlling the AUV’s heading and controlling the AUV to a desired distance from the tunnel wall. To be able to evaluate the performance of the system, a graphical user interface (GUI) has been developed. The GUI enables the operator to change control settings during simulations. Two proposed control strategies are presented with simulated results. / Syftet med examensarbetet är att undersöka hur en autonom undervattensfarkost (AUV) bör agera i en undervattenstunnel miljö. Avhandlingen föreslår sensorer, reglerstrategier, uppdragsbeskrivning med mera som krävs för tunneluppdrag. En beteendebaserad (behavior-based) reglerstrategi har utvecklats för att styra AUV:n. Reglerstrategin används i mellersta nivån i farkostens reglering, det vill säga den reaktiva regleringen som beskriver hur farkosten ska styra genom en tunnel samtidigt som andra händelser beaktas. Reglerstrategin har även delats upp i två delar: reglering av AUV:ns kurs och reglering av AUV:n till ett önskat avstånd från tunnelns vägg. För att kunna verifiera funktionaliteten av systemet så har även ett grafiskt användargränssnitt utvecklats. Gränssnittet möjliggör att man kan ändra reglerparametrar under en simulering. Två föreslagna reglerstrategier presenteras med tillhörande resultat.

­^¤åºK­n¡G In today¡¦s knowledge economy age, competition emphasizes on human capital and knowledge accumulation. It¡¦s quite a different way from manufacturing era when competitive competencies mainly come from machine assets, materials and land capital. Therefore, high quality of human resources becomes one of the key sources of business success and competitive advantages. Because the process of service delivery needs high degree of personal contact, the quality of service providers has great impacts on the quality of service provided. And cultivating employees with service intent and customer orientation is an important mission which can¡¦t be waited. The purpose of this study is to develop a scale using to measure the degree of internal marketing behaviors by referring to researches in the past and asking opinions from academic and practical experts. This study still tries to study the securities salesmen in the southern branches of integrated securities corporations and verify the relationship between the degree of internal marketing and behavior- based evaluation policies perceived by securities salesmen and the degree of their job satisfaction and customer orientation. In addition, this study also tries to test if securities salesmen feel higher degree of satisfaction about internal marketing policy, they have higher degree of job satisfaction and perform higher degree of customer orientation. There are six findings of this study. They are as the following: (1) the perceived degree of internal marketing and behavior- based evaluation policies and the satisfaction degree of internal marketing policy have positive impacts on the degree of security salesmen¡¦s job satisfaction. (2) job satisfaction and customer orientation isn¡¦t significantly related, and there could be some other factors which have impacts on this relationship. (3) the perceived degrees of internal marketing and behavior- based evaluation policies have positive impacts on the degree of securities salesmen¡¦s customer orientation. (4) the satisfaction degree of internal marketing policy is negatively related to securities salesmen¡¦s customer orientation. (5) the perceived degree of behavior- based evaluation policy moderates the relationship of the perceived degree of internal marketing policy and customer orientation. (6) the perceived degree of behavior- based evaluation policy moderates the relationship of the satisfaction degree of internal marketing policy and customer orientation. And the outcome of factor analysis of the perceived degree of internal marketing policy is six factor constructs, compensation, personal growth, comprehension, supervisors¡¦ concern, promotion and place constructs. The original product constructs is divided into three constructs, compensation, personal growth and comprehension, and the other constructs such as promotion, supervisors¡¦ concern and place constructs remain the same. The constructs we gain from this study are consistent with marketing mix, 7Ps.