Silvio do Lago Pereira

A logic-based agent that plans for extended reachability goals

Planning to reach a goal is an essential capability for rational agents. In general, a goal specifies a condition to be achieved at the end of the plan execution. In this article, we introduce nondeterministic planning for extended reachability goals (i.e., goals that also specify a condition to be preserved during the plan execution). We show that, when this kind of goal is considered, the temporal logic ctl turns out to be inadequate to formalize plan synthesis and plan validation algorithms. This is mainly due to the fact that the ctl’s semantics cannot discern among the various actions that produce state transitions. To overcome this limitation, we propose a new temporal logic called α-ctl. Then, based on this new logic, we implement a planner capable of synthesizing reliable plans for extended reachability goals, as a side effect of model checking.

Planning with Abduction: A Logical Framework to Explore Extensions to Classical Planning

In this work we show how a planner implemented as an abductive reasoning process can have the same performance and behavior as classical planning algorithms. We demonstrate this result by considering three different versions of an abductive event calculus planner on reproducing some important comparative analyses of planning algorithms found in the literature. We argue that a logic-based planner, defined as the application of general purpose theorem proving techniques to a general purpose action formalism, can be a very solid base for the research on extending the classical planning approach.

Strong Probabilistic Planning

We consider the problem of synthesizing policies, in domains where actions have probabilisticeffects, that are optimal in the expected-caseamong the optimal worst-case strongpolicies. Thus we combine features from nondeterministic and probabilistic planning in a single framework. We present an algorithm that combines dynamic programming and model checking techniques to find plans satisfying the problem requirements: the strong preimage computation from model checking is used to avoid actions that lead to cycles or dead ends, reducing the model to a Markov Decision Process where all possible policies are strong and worst-case optimal (i.e., successful and minimum length with probability 1). We show that backward induction can then be used to select a policy in this reduced model. The resulting algorithm is presented in two versions (enumerative and symbolic); we show that the latter version allows planning with extended reachability goals.

System design modification with actions

System designers are expected to use error-detecting and correcting techniques. Although, model checking approaches have been used for verification of errors in large complex systems, they can only detect the error. the task of correcting the system design (called model update) is completely left to the system designer. Recent works on model update can suggest changes in the system model which do not consider domain contingencies and constraints. In this paper, we present a model update approach that can be used to automatically suggest modifications in a system based on the actions that are behind state transitions and a set of domain constraints. We claim that with this approach we can develop more realistic system error-correcting tools.

Using α-ctl to Specify Complex Planning Goals

The temporal logic ctl has been the preferred specificationlanguage in the model checking framework. However, when thisframework is used for nondeterministic planning, it is not adequateto deal with many useful planning problems with temporally extendedgoals. This is because the validity of ctl formulas expressing suchgoals is not evaluated on the planning domain, but on the executionstructure of the policy synthesized by the planning algorithm. Inprevious work we have presented a new variant of ctl, namedαctl, which semantics can be defined directly onthe planning domain. An advantage of this new logic is that plansynthesis can be obtained as a collateral effect of verifying thevalidity of a formula in the planning domain. In this paper we showhow to use α-ctl to express some complex planninggoals.

High-Level Robot Programming: An Abductive Approach Using Event Calculus

This paper proposes a new language that can be used to build high-level robot controllers with high-level cognitive functions such as plan specification, plan generation, plan execution, perception, goal formulation, communication and collaboration. The proposed language is based on Golog, a language that uses the situation calculus as a formalism to describe actions and deduction as an inference rule to synthesize plans. On the other hand, instead of situation calculus and deduction, the new language uses event calculus and abductive reasoning to synthesize plans. As we can forsee, this change of paradigm allows the agent to reason about partial order plans, making possible a more flexible integration between deliberative and reactive behaviors.

A planner agent that tries its best in presence of nondeterminism

Abstract In many nondeterministic planning domains, an agent whose goal is to achieve φ may not succeed in finding a policy that can guarantee that all paths from the initial state lead to a final state where φ holds (strong solution). Nevertheless, if the agent is trying its best to achieve φ, it cannot give up. Instead, it may be inclined to accept weaker guarantees, such as having a path leading to φ from any intermediate state reached by the policy (strong-cyclic solution), or even less, such as having at least one path leading to φ from the initial state (weak solution). But the agent should choose among such different options based on their availability in each situation. Although the specification of this type of goal has been addressed before, in this paper we show how a planner based on the branching time temporal logic α- ctl can be used to plan for intuitive and useful goals of the form “try your best to achieve φ”.