Collaboration rules for autonomous software agents

Can autonomous software agents that are distributed over a computer network collaborate effectively? Both empirical evidence and theory suggest that they can. Moreover, there seem to be simple rules for designing problem-solving organizations in which collaboration among such agents is automatic and scale-effective (adding agents tends to improve solution-quality; adding computers tends to improve solution-speed). This paper develops some of these rules for off-line problems and argues that they can be extended for the on-line (real-time) control of power systems.     

自动驾驶软件测试技术研究综述

自动驾驶系统（autonomous driving system,ADS）是一种集成高精度传感器、人工智能和地图导航系统等模块的信息—物理融合系统。该类系统中的自动驾驶软件完成了从高级辅助驾驶到无人驾驶任务中关键的感知、定位、预测、规划和控制任务。随着深度学习和强化学习等人工智能技术的发展和车载硬件设备的不断升级,高级别的自动驾驶软件已经逐渐应用于多种安全攸关的场景中,保障其运行稳定性与可靠性的测试技术逐渐成为学术界和产业界的研究重点。本文在广泛调研国内外文献基础上,对自动驾驶软件测试技术进行了深入分析与梳理。结合自动驾驶软件的架构特点及系统特征,讨论了面向自动驾驶系统的仿真测试和实景测试,以及面向组件的测试技术。其中,在仿真方法方面,分析了软件仿真、半实体仿真和在环仿真等技术;在仿真对象方面,讨论了静态环境仿真、动态场景仿真、传感器仿真和车辆动力学仿真等。同时,本文介绍了当前实景测试的进展与情况,重点分析了实景测试案例中的得失优劣。在面向自动驾驶软件组件的测试技术方面,重点讨论了当前数据驱动技术在感知组件、决策规划组件,以及控制组件测试方面的进展。最后,本文总结分析了自动驾驶软件测试当前面临的挑战,并对未来自动驾驶软件测试技术的研究方向和研究重点进行了展望。     

Evolutionary testing of software with function-assigned flags

Evolutionary structural testing, an approach to automatically generate relevant unit test data, encounters difficulties when the software being tested contains boolean variables. This issue, known as the flag problem, has been studied by many researchers. However, previous work does not address the issue of function-assigned flags which constitutes a special type of flag problem that often occurs in the context of object-orientation. This paper elaborates on a new approach to the flag problem that can also handle function-assigned flags while being applicable to the conventional flag problem, as well. It relies on a code transformation that leads to an improved fitness landscape which provides better guidance to the evolutionary search. We present seven case studies including a fitness landscape analysis and experimental results. The results show that the suggested code transformation improves evolutionary structural testing in the presence of function-assigned flags.     

Automated assembly of Internet-scale software systems involving autonomous agents

On the Internet, there is a great amount of distributed software entities deployed independently and behaving autonomously. This paper describes an automated approach to constructing Internet-scale software systems based on autonomous software agents. In the approach, the systems are modeled by interconnected divisions and cooperative roles. The approach adopts a dynamic trial-and-evaluation strategy to select high quality autonomous agents to undertake the responsibilities of roles, and implements a special mobile agent, called delegate, carrying the interaction information specified for responsibilities of roles to facilitate the interoperations among autonomous agents. The experiments show that the approach is highly scalable and improves the overall qualities of systems remarkably.     

Specification and testing of autonomous agents in e‐commerce systems

This paper presents a generic formal framework to specify and test autonomous e‐commerce agents. First, the formalism to represent the behaviour of agents is introduced. The corresponding machinery to define how implementations can be tested follows. Two testing approaches are considered. The first of them, which can be called active, is based on stimulating the implementation under test (IUT) with a test. The peculiarity is that tests will be defined as a special case of autonomous e‐commerce agent. The second approach, which can be called passive, consists of observing the behaviour of the tested agent in an environment containing other agents. As a case study the framework is applied to the e‐commerce system Kasbah. Copyright © 2005 John Wiley & Sons, Ltd.     

Resource-sharing among autonomous agents

We study a scenario for cloud services based on autonomous resource management agents in situations of competition for limited resources. In the scenario, autonomous agents make independent decisions on resource consumption in a competitive environment. Altruistic and selfish strategies for agent behaviour are simulated and compared with respect to whether they lead to successful resource management in the overall system, and how much information exchange is needed among the agents for the strategies to work. Our results imply that local agent information could be sufficient for global optimisation. Also, the selfish strategy proved stable compared to uninformed altruistic behaviour.     

Evolutionary mutation testing

Context

Mutation testing is a testing technique that has been applied successfully to several programming languages. However, it is often regarded as computationally expensive, so several refinements have been proposed to reduce its cost. Moreover, WS-BPEL compositions are being widely adopted by developers, but present new challenges for testing, since they can take much longer to run than traditional programs of the same size. Therefore, it is interesting to reduce the number of mutants required.

Objective

We present Evolutionary Mutation Testing (EMT), a novel mutant reduction technique for finding mutants that help derive new test cases that improve the quality of the initial test suite. It uses evolutionary algorithms to reduce the number of mutants that are generated and executed with respect to the exhaustive execution of all possible mutants, keeping as many difficult to kill and potentially equivalent mutants (strong mutants) as possible in the reduced set.

Method

To evaluate EMT we have developed GAmera, a mutation testing system powered by a co-evolutive genetic algorithm. We have applied this system to three WS-BPEL compositions to estimate its effectiveness, comparing it with random selection.

Results

The results obtained experimentally show that EMT can select all strong mutants generating 15% less mutants than random selection in over 20% less time for complex compositions. When generating a percentage of all mutants, EMT finds on average more strong mutants than random selection. This has been confirmed to be statistically significant within a 99.9% confidence interval.

Conclusions

EMT has reduced for the three tested compositions the number of mutants required to select those which are useful to derive new test cases that improve the quality of the test suite. The directed search performed by EMT makes it more effective than random selection, especially as compositions become more complex and the search space widens.     

Evolutionary functional testing

The development and testing of software-based systems is an essential activity for the automotive industry. The 50–70 software-based systems with different complexities and developed by various suppliers are installed in today's premium vehicles, communicating with each other via different bus systems. The integration and testing of systems of this complexity is a very challenging task. The aim of testing is to detect faults in the systems under test and to convey confidence in the correct functioning of the systems if no faults are found during comprehensive testing. Faults not found in the different testing phases could have significant consequences that range from customer dissatisfaction to damage of physical property or, in safety-relevant areas, even to the endangering of human lives. Therefore, the thorough testing of developed systems is essential. Evolutionary testing tries to improve the effectiveness and efficiency of the testing process by transforming testing objectives into search problems, and applying evolutionary computation in order to solve them. The most important class of testing methods is functional testing. However, functional testing is difficult to automate by evolutionary testing. This work will describe how evolutionary testing could be applied to automate functional testing in general and the testing of complex automotive systems in particular. It presents two case studies and shows how evolutionary testing is effective at finding faults in the functional behaviour of these systems. In addition, a quantitative comparison with manual and random test case selection is done for one application.     

Chaotic neurodynamics for autonomous agents

Mesoscopic level neurodynamics study the collective dynamical behavior of neural populations. Such models are becoming increasingly important in understanding large-scale brain processes. Brains exhibit aperiodic oscillations with a much more rich dynamical behavior than fixed-point and limit-cycle approximation allow. Here we present a discretized model inspired by Freeman's K-set mesoscopic level population model. We show that this version is capable of replicating the important principles of aperiodic/chaotic neurodynamics while being fast enough for use in real-time autonomous agent applications. This simplification of the K model provides many advantages not only in terms of efficiency but in simplicity and its ability to be analyzed in terms of its dynamical properties. We study the discrete version using a multilayer, highly recurrent model of the neural architecture of perceptual brain areas. We use this architecture to develop example action selection mechanisms in an autonomous agent.     

Normalized Interactions between autonomous agents

The CEC Project GOAL (Esprit 6283) aims to develop generic software tools to support a new project management paradigm, in which projects are collaborative, decentralised and inter-organizational. To support inter-organizational interaction, communication and cooperation, we are developing a design framework for formalizing the flow of information between organizations, specifying access to and provision of project services, and defining project-wide standards and procedures. This framework is based on normalizing interactions between autonomous software agents by specifying messages and protocols for inter-agent communication and cooperation. This paper reviews the framework, and then focusses on the specification and implementation of a case study, the automation of a distributed document review procedure. This is both a successful proof of concept and a demonstration of how Artificial Intelligence technologies can support inter-organizational project management. It also points the way to agent brokering, an enhancement of object brokering in distributed open systems, where the satisfaction of service requests can be subject to negotiation.Supported by CEC Esprit Project GOAL (Esprit 6283) and CEC Esprit BRA Medlar II (Esprit 6471).Supported by CEC Esprit Project GOAL (Esprit 6283).     

Simulating belief systems of autonomous agents

Autonomous agents in computer simulations do not have the usual mechanisms to acquire information as do their human counterparts. In many such simulations, it is not desirable that the agent have access to complete and correct information about its environment. We examine how imperfection in available information may be simulated in the case of autonomous agents. We determine probabilistically what the agent may detect, through hypothetical sensors, in a given situation. These detections are combined with the agent's knowledge base to infer observations and beliefs. Inherent in this task is a degree of uncertainty in choosing the most appropriate observation or belief. We describe and compare two approaches — a numerical approach and one based on defeasible logic — for simulating an appropriate belief in light of conflicting detection values at a given point in time. We discuss the application of this technique to autonomous forces in combat simulation systems.     

Stochastic software testing

This paper presents a method for test case selection that allows a formal approach to testing software. The two main ideas are (1) that testers create stochastic models of software behavior instead of crafting individual test cases and (2) that specific test cases are generated from the stochastic models and applied to the software under test. This paper describes a method for creating a stochastic model in the context of a solved example. We concentrate on Markov models and show how non‐Markovian behavior can be embedded in such models without violating the Markov property.     

A new approach in development of distributed framework for automated software testing using agents

Software testing is the technical kernel of software quality engineering, and to develop critical and complex software systems not only requires a complete, consistent and unambiguous design, and implementation methods, but also a suitable testing environment that meets certain requirements, particularly, to face the complexity issues. Traditional methods, such as analyzing each requirement and developing test cases to verify correct implementation, are not effective in understanding the software’s overall complex behavior. In that respect, existing approaches to software testing are viewed as time-consuming and insufficient for the dynamism of the modern business environment. This dynamics requires new tools and techniques, which can be employed in tandem with innovative approaches to using and combining existing software engineering methods. This work advocates the use of a recently proposed software engineering paradigm, which is particularly suited to the construction of complex and distributed software-testing systems, which is known as Agent-Oriented Software Engineering. This methodology is a new one, which gives the basic approach to agent-based frameworks for testing.     

Conformance testing of software

People who buy software want a guarantee that it works. When they cannot satisfy themselves, and do not wholly trust the programmer, it is natural to seek the help of a third party who will certify the software. So far, formal certification is available only for compilers. This paper discusses some of the theoretical problems and reviews existing compiler validation schemes. Finally there is a brief examination of other methods of quality assurance that might provide a better solution.     

Argumentation-based negotiation planning for autonomous agents

When we negotiate, the arguments uttered to persuade the opponent are not the result of an isolated analysis, but of an integral view of the problem that we want to agree about. Before the negotiation starts, we have in mind what arguments we can utter, what opponent we can persuade, which negotiation can finish successfully and which cannot. Thus, we plan the negotiation, and in particular, the argumentation. This fact allows us to take decisions in advance and to start the negotiation more confidently. With this in mind, we claim that this planning can be exploited by an autonomous agent. Agents plan the actions that they should execute to achieve their goals. In these plans, some actions are under the agent's control, while some others are not. The latter must be negotiated with other agents. Negotiation is usually carried out during the plan execution. In our opinion, however, negotiation can be considered during the planning stage, as in real life. In this paper, we present a novel approach to integrate argumentation-based negotiation planning into the general planning process of an autonomous agent. This integration allows the agent to take key decisions in advance. We evaluated this proposal in a multiagent scenario by comparing the performance of agents that plan the argumentation and agents that do not. These evaluations demonstrated that performance improves when the argumentation is planned, specially, when the negotiation alternatives increase.     

Intermediating user-DSS interaction with autonomous agents

The use of advanced decision support system (DSS) capabilities is hampered by the inadequacy of a "toolbox" organization of DSS from the user's perspective. In such a setup, the user is assumed to have all the knowledge and skills necessary to appropriately use the tools provided by the system in the decision-making process. This paper proposes a model for the use of autonomous agents as intermediaries between the users and the system. The model is organized around the human problem-solving process. The paper elaborates on the types of intermediary agents and the architecture for a DSS. The approach is illustrated using the prototype for an investment DSS.     

Socially intelligent reasoning for autonomous agents

Socially intelligent agents are autonomous problem solvers that have to achieve their objectives by interacting with other similarly autonomous entities. A major concern, therefore, is with the design of the decision-making mechanism that such agents employ in order to determine which actions to take to achieve their goals. We propose a framework for making socially acceptable decisions, based on social welfare functions, that combines social and individual perspectives in a unified and flexible manner. The framework is realized in an exemplar computational setting and an empirical analysis is made of the relative performance of varying sociable decision-making functions in a range of environments. This analysis is then used to design an agent that adapts its decision-making to reflect the resource constraints that it faces at any given time. A further round of empirical evaluation shows how adding such a meta-level mechanism enhances the performance of the agent by directing reasoning to adopt different strategies in different contexts. Finally, the possibility and efficacy of making the metalevel mechanism adaptive, so that experience of past encounters can be factored into the decision-making, is demonstrated     

Negotiation decision functions for autonomous agents

We present a formal model of negotiation between autonomous agents. The purpose of the negotiation is to reach an agreement about the provision of a service by one agent for another. The model defines a range of strategies and tactics that agents can employ to generate initial offers, evaluate proposals and offer counter proposals. The model is based on computationally tractable assumptions, demonstrated in the domain of business process management and empirically evaluated.     

Dynamic clustering of maps in autonomous agents

The problem of organizing and exploiting spatial knowledge for navigation is an important issue in the field of autonomous mobile systems. In particular, partitioning the environment map into connected clusters allows for significant topological features to be captured and enables decomposition of path-planning tasks through a divide-and-conquer policy. Clustering by discovery is a procedure for identifying clusters in a map being learned by exploration as the agent moves within the environment, and yields a valid clustering of the available knowledge at each exploration step. In this work, we define a fitness measure for clustering and propose two incremental heuristic algorithms to maximize it. Both algorithms determine clusters dynamically according to a set of topological and metric criteria. The first one is aimed at locally minimizing a measure of “scattering” of the entities belonging to clusters, and partially rearranges the existing clusters at each exploration step. The second estimates the positions and dimensions of clusters according to a global map of density. The two algorithms are compared in terms of optimality, efficiency, robustness, and stability