anmin Ji

Toward open knowledge enabling for human-robot interaction

This paper presents an effort to enable robots to utilize open-source knowledge resources autonomously for human-robot interaction. The main challenges include how to extract knowledge in semi-structured and unstructured natural languages, how to make use of multiple types of knowledge in decision making, and how to identify the knowledge that is missing. A set of techniques for multi-mode natural language processing, integrated decision making, and open knowledge searching is proposed. The OK-KeJia robot prototype is implemented and evaluated, with special attention to two tests on 11,615 user tasks and 467 user desires. The experiments show that the overall performance improves remarkably due to the use of appropriate open knowledge.

Integrating NLP with Reasoning about Actions for Autonomous Agents Communicating with Humans

We present a first effort to integrate NLP with ASP for autonomous agents, especially service robots, communicating with humans. We implemented a prototype system and tested it in a simple home environment, which demonstrated the feasibility of our approach under current settings and the possibility that our approach will benefit from future advancement of research on NLP, ASP, and related areas.

Computing Loops with at Most One External Support Rule

A consequence of a logic program under answer set semantics is one that is true for all answer sets. This article considers using loop formulas to compute some of these consequences in order to increase the efficiency of answer set solvers. Since computing loop formulas are in general intractable, we consider only loops with either no external support or at most one external support, as their loop formulas are either unit or binary clauses. We show that for disjunctive logic programs, loop formulas of loops with no external support can be computed in polynomial time, and that an iterative procedure using unit propagation on these formulas and the program completion computes the well-founded models in the case of normal logic programs and the least fixed point of a simplification operator used by DLV for disjunctive logic programs. For loops with at most one external support, their loop formulas can be computed in polynomial time for normal logic programs, but are NP-hard for disjunctive programs. So for normal logic programs, we have a procedure similar to the iterative one for loops without any external support, but for disjunctive logic programs, we present a polynomial approximation algorithm. All these algorithms have been implemented, and our experiments show that for certain logic programs, the consequences computed by our algorithms can significantly speed up current ASP solvers cmodels, clasp, and DLV.

Computing Loops with at Most One External Support Rule for Disjunctive Logic Programs

We extend to disjunctive logic programs our previous work on computing loop formulas of loops with at most one external support. We show that for these logic programs, loop formulas of loops with no external support can be computed in polynomial time, and if the given program has no constraints, an iterative procedure based on these formulas, the program completion, and unit propagation computes the least fixed point of a simplification operator used by DLV. We also relate loops with no external supports to the unfounded sets and the well-founded semantics of disjunctive logic programs by Wang and Zhou. However, the problem of computing loop formulas of loops with at most one external support rule is NP-hard for disjunctive logic programs. We thus propose a polynomial algorithm for computing some of these loop formulas, and show experimentally that this polynomial approximation algorithm can be effective in practice.

From turner's logic of universal causation to the logic of GK

Logic of knowledge and justified assumptions, also known as logic of grounded knowledge (GK), was proposed by Lin and Shoham as a general logic for nonmonotonic reasoning. To date, it has been used to embed in it default logic, autoepistemic logic, and general logic programming under stable model semantics. Besides showing the generality of GK as a logic for nonmonotonic reasoning, these embeddings shed light on the relationships among these other logics. Along this line, we show that Turners logic of universal causation can be naturally embedded into logic of GK as well.

A weighted causal theory for acquiring and utilizing open knowledge

Motivated by enabling intelligent robots/agents to take advantage of open-source knowledge resources to solve open-ended tasks, a weighted causal theory is introduced as the formal basis for the development of these robots/agents. The action model of a robot/agent is specified as a causal theory following McCain and Turners nonmonotonic causal theories. New knowledge is needed when the robot/agent is given a user task that cannot be accomplished only with the action model. This problem is cast as a variant of abduction, that is, to find the most suitable set of causal rules from open-source knowledge resources, so that a plan for accomplishing the task can be computed using the action model together with the acquired knowledge. The core part of our theory is constructed based on credulous reasoning and the complexity of corresponding abductive reasoning is analyzed. The entire theory is established by adding weights to hypothetical causal rules and using them to compare competing explanations which induce causal models satisfying the task. Moreover, we sketch a model theoretic semantics for the weighted causal theory and present an algorithm for computing a weighted-abductive explanation. An application of the techniques proposed in this paper is illustrated in an example on our service robot, KeJia, in which the robot tries to acquire proper knowledge from OMICS, a large-scale open-source knowledge resource, and solve new tasks with the knowledge. The work aims to provide a formal basis for acquiring and utilizing open knowledge.The core concept of knowledge gap is formulized based on causal theories.The weighted knowledge gap is introduced, providing insights for the expected applications.The complexity results of related reasoning tasks are presented thoroughly.The application of the theoretical results is illustrated in an example on a real robot.

Towards Metareasoning for Human-Robot Interaction

This paper proposes a model of metareasoning for Human-Robot Interaction (HRI). Robots’ basic abilities for HRI—planning, learning and dialogue—are characterized as three loops in the model, with each spanning ground, object and meta-level. The model provides a conceptualization of HRI and a framework for incremental development of large HRI systems such as service robots by building meta-level functions on top of existing ground/object level components. A case-study focusing on meta-level control shows that the approach is effective and efficient for some application domains. In particular, meta-level control suggests a new opportunity to speed up planning while preserving completeness without any change to object level planners. The experiments also show that, for some basic HRI tasks, there are simple meta-level strategies with performances better than the common strategy in previous work.

Turner's Logic of Universal Causation, Propositional Logic, and Logic Programming

Turners logic of universal causation is a general logic for nonmonotonic reasoning. It has its origin in McCain and Turners causal action theories which have been translated to propositional logic and logic programming with nested expressions. In this paper, we propose to do the same for Turners logic, and show thatTurners logic can actually be mapped to McCain and Turners causal theories. These results can be used to construct a system for reasoning in Turners logic.

Simulation Competitions on Domestic Robots

This paper reports a series of simulation competitions on domestic robots. All of these five competitions were based on a simulation platform focused on evaluating high-level functions of a domestic robot, including task planning and dialogue understanding. The object of holding these competitions is to promote research and development of service robots while avoiding limitations imposed by hardware of real robots. We also analyze the results and performances of participating teams since the competition was first held in 2009, showing that more and more terms are participating and they are performing better and better.

Help Me! Sharing of Instructions Between Remote and Heterogeneous Robots

Service robots frequently face similar tasks. However, they are still not able to share their knowledge efficiently on how to accomplish those tasks. We introduce a new framework, which allows remote and heterogeneous robots to share instructions on the tasks assigned to them. This framework is used to initiate tasks for the robots, to receive or provide instructions on how to accomplish the tasks, and to ground the instructions in the robots’ capabilities. We demonstrate the feasibility of the framework with experiments between two geographically distributed robots and analyze the performance of the proposed framework quantitatively.