M. Birna van Riemsdijk

A Programming Language for Cognitive Agents: Goal-Directed 3APL

This paper presents the specification of a programming language for cognitive agents. This programming language is an extension of 3APL (An Abstract Agent Programming Language) and allows the programmer to implement agents’ mental attitudes like beliefs, goals, plans, and actions, and agents’ reasoning rules by means of which agents can modify their mental attitudes. The formal syntax and semantics of this language is presented as well as a discussion on the deliberation cycle and an example.

Shared Mental Models

The notion of a shared mental model is well known in the literature regarding team work among humans. It has been used to explain team functioning. The idea is that team performance improves if team members have a shared understanding of the task that is to be performed and of the involved team work. We maintain that the notion of shared mental model is not only highly relevant in the context of human teams, but also for teams of agents and for human-agent teams. However, before we can start investigating how to engineer agents on the basis of the notion of shared mental model, we first have to get a better understanding of the notion, which is the aim of this paper. We do this by investigating which concepts are relevant for shared mental models, and modeling how they are related by means of UML. Through this, we obtain a mental model ontology. Then, we formally define the notion of shared mental model and related notions. We illustrate our definitions by means of an example.

Semantics of declarative goals in agent programming

This paper addresses the notion of declarative goals as used in agent programming. Declarative goals describe desirable states, and semantics of these goals in an agent programming context can be defined in various ways. This paper defines two semantics for goals, with one based on default logic. The semantics are partly motivated by an analysis of other proposals that have been done in the literature. Further, we establish relations between and properties of these semantics.

Coactive design: designing support for interdependence in joint activity

Coactive Design is a new approach to address the increasingly sophisticated roles that people and robots play as the use of robots expands into new, complex domains. The approach is motivated by the desire for robots to perform less like teleoperated tools or independent automatons and more like interdependent teammates. In this article, we describe what it means to be interdependent, why this is important, and the design implications that follow from this perspective. We argue for a human-robot system model that supports interdependence through careful attention to requirements for observability, predictability, and directability. We present a Coactive Design method and show how it can be a useful approach for developers trying to understand how to translate high-level teamwork concepts into reusable control algorithms, interface elements, and behaviors that enable robots to fulfill their envisioned role as teammates. As an example of the coactive design approach, we present our results from the DARPA Virtual Robotics Challenge, a competition designed to spur development of advanced robots that can assist humans in recovering from natural and man-made disasters. Twenty-six teams from eight countries competed in three different tasks providing an excellent evaluation of the relative effectiveness of different approaches to human-machine system design.

Goal types in agent programming

This paper discusses three types of declarative goals and motivates their integration in logic-based agent-oriented programming languages. These goal types are perform goals, achieve goals, and maintain goals. A goal type is considered as a specific agent attitude towards goals. The semantics for each goal type is explained from an operational perspective. It is argued that the suggested semantics of the goal types ensure some desirable and expected properties.

Joint Activity Testbed: Blocks World for Teams (BW4T)

This demonstration will be the presentation of a new testbed for joint activity. The domain for this demonstration will be similar to the classic AI planning problem of Blocks World (BW) extended into what we are calling Blocks World for Teams (BW4T). By teams, we mean at least two, but usually more members. Additionally, we do not restrict the membership to artificial agents, but include and in fact expect human team members. Study of joint activity of heterogeneous teams is the main function of the BW4T testbed.

Goal-oriented modularity in agent programming

Modularization is widely recognized as a central issue in software engineering. In this paper we address the issue of modularization in cognitive agent programming languages. We discuss existing approaches to modularity in cognitive agent programming. Then, we propose a new kind of modularity, i.e., goal-oriented modularity, which takes the goals of an agent as the basis for modularization. Further, we present a formal semantics of goal-oriented modularity in the context of the 3APL agent programming language.

Goals in conflict: semantic foundations of goals in agent programming

This paper addresses the notion of (declarative) goals as used in agent programming. Goals describe desirable states, and semantics of these goals in an agent programming context can be defined in various ways. We focus in this paper on the representation of conflicting goals. In particular, we define two semantics for goals, one for unconditional goals and one for conditional goals. The first is based on propositional logic, and the latter is based on default logic. We establish relations between and properties of these semantics.

Satisfying maintenance goals

A rational agent derives its choice of action from its beliefs and goals. Goals can be distinguished into achievement goals and maintenance goals. The aim of this paper is to define a mechanism which ensures the satisfaction of maintenance goals. We argue that such a mechanism requires the agent to look ahead, in order to make sure that the execution of actions does not lead to a violation of a maintenance goal. That is, maintenance goals may constrain the agent in choosing its actions. We propose a formal semantics of maintenance goals based on the notion of lookahead, and analyze the semantics by proving some properties. Additionally, we discuss the issue of achievement goal revision, in case the maintenance goals are so restrictive that all courses of action for satisfying achievement goals will lead to a violation of maintenance goals.

Dynamics of declarative goals in agent programming

In this paper, the notion of declarative goals as used in agent programming is central. Declarative goals describe desirable states and are updated during the execution of an agent. These goal dynamics are analyzed by distinguishing and formalizing various notions of goal dropping and goal adoption. Furthermore, possible motivations for an agent to drop or adopt goals are identified. Based on these motivations, we define specific mechanisms for implementing dropping and adoption. We show how these mechanisms are related to the general definitions of dropping and adoption.