Esra Erdem

Combining high-level causal reasoning with low-level geometric reasoning and motion planning for robotic manipulation

We present a formal framework that combines high-level representation and causality-based reasoning with low-level geometric reasoning and motion planning. The frame-work features bilateral interaction between task and motion planning, and embeds geometric reasoning in causal reasoning, thanks to several advantages inherited from its underlying components. In particular, our choice of using a causality-based high-level formalism for describing action domains allows us to represent ramifications and state/transition constraints, and embed in such formal domain descriptions externally defined functions implemented in some programming language (e.g., C++). Moreover, given such a domain description, the causal reasoner based on this formalism (i.e., the Causal Calculator) allows us to compute optimal solutions (e.g., shortest plans) for elaborate planning/prediction problems with temporal constraints. Utilizing these features of high-level representation and reasoning, we can combine causal reasoning, motion planning and geometric planning to find feasible kinematic solutions to task-level problems. In our framework, the causal reasoner guides the motion planner by finding an optimal task-plan; if there is no feasible kinematic solution for that task-plan then the motion planner guides the causal reasoner by modifying the planning problem with new temporal constraints. Furthermore, while computing a task-plan, the causal reasoner takes into account geometric models and kinematic relations by means of external predicates implemented for geometric reasoning (e.g., to check some collisions); in that sense the geometric reasoner guides the causal reasoner to find feasible kinematic solutions. We illustrate an application of this framework to robotic manipulation, with two pantograph robots on a complex assembly task that requires concurrent execution of actions. A short video of this application accompanies the paper.

Answer set programming for collaborative housekeeping robotics: representation, reasoning, and execution

Answer set programming (ASP) is a knowledge representation and reasoning paradigm with high-level expressive logic-based formalism, and efficient solvers; it is applied to solve hard problems in various domains, such as systems biology, wire routing, and space shuttle control. In this paper, we present an application of ASP to housekeeping robotics. We show how the following problems are addressed using computational methods/tools of ASP: (1) embedding commonsense knowledge automatically extracted from the commonsense knowledge base ConceptNet, into high-level representation, and (2) embedding (continuous) geometric reasoning and temporal reasoning about durations of actions, into (discrete) high-level reasoning. We introduce a planning and monitoring algorithm for safe execution of plans, so that robots can recover from plan failures due to collision with movable objects whose presence and location are not known in advance or due to heavy objects that cannot be lifted alone. Some of the recoveries require collaboration of robots. We illustrate the applicability of ASP on several housekeeping robotics problems, and report on the computational efficiency in terms of CPU time and memory.

Inferring Phylogenetic Trees Using Answer Set Programming

We describe the reconstruction of a phylogeny for a set of taxa, with a character-based cladistics approach, in a declarative knowledge representation formalism, and show how to use computational methods of answer set programming to generate conjectures about the evolution of the given taxa. We have applied this computational method in two domains: historical analysis of languages and historical analysis of parasite-host systems. In particular, using this method, we have computed some plausible phylogenies for Chinese dialects, for Indo-European language groups, and for Alcataenia species. Some of these plausible phylogenies are different from the ones computed by other software. Using this method, we can easily describe domain-specific information (e.g., temporal and geographical constraints), and thus prevent the reconstruction of some phylogenies that are not plausible.

Temporal phylogenetic networks and logic programming

The concept of a temporal phylogenetic network is a mathematical model of evolution of a family of natural languages. It takes into account the fact that languages can trade their characteristics with each other when linguistic communities are in contact, and also that a contact is only possible when the languages are spoken at the same time. We show how computational methods of answer set programming and constraint logic programming can be used to generate plausible conjectures about contacts between prehistoric linguistic communities, and illustrate our approach by applying it to the evolutionary history of Indo-European languages.

Applications of Answer Set Programming

ASP has been applied fruitfully to a wide range of areas in AI and in other fields, both in academia and in industry, thanks to the expressive representation languages of ASP and the continuous improvement of ASP solvers. We present some of these ASP applications, in particular, in knowledge representation and reasoning, robotics, bioinformatics and computational biology as well as some industrial applications. We discuss the challenges addressed by ASP in these applications and emphasize the strengths of ASP as a useful AI paradigm.

Bridging the Gap between High-Level Reasoning and Low-Level Control

We present a formal framework where a nonmonotonic formalism (the action description language

Rectilinear Steiner Tree Construction Using Answer Set Programming

{\cal C}+

Wire Routing and Satisfiability Planning

) is used to provide robots with high-level reasoning, such as planning, in the style of cognitive robotics. In particular, we introduce a novel method that bridges the high-level discrete action planning and the low-level continuous behavior by trajectory planning. We show the applicability of this framework on two LEGO MINDSTORMS NXT robots, in an action domain that involves concurrent execution of actions that cannot be serialized.

Geometric rearrangement of multiple movable objects on cluttered surfaces: A hybrid reasoning approach

We introduce a new method for Rectilinear Steiner Tree (RST) construction in a graph, using answer set programming. This method provides a formal representation of the problem as a logic program whose answer sets correspond to solutions. The answer sets for a logic program can be computed by special systems called answer set solvers. We describe the method for RST construction in the context of VLSI routing where multiple pins in a given placement of a chip are connected by an RST. Our method is different from the existing methods mainly in three ways. First, it always correctly determines whether a given RST routing problem is solvable, and it always produces a solution if one exists. Second, some enhancements of the basic problem, in which lengths of wires connecting the source pin to sink pins are restricted, can be easily represented by adding some rules. Our method guarantees to find a tree if one exists, even when the total wire length is not minimum. Third, routing problems with the presence of obstacles can be solved. With this approach, we have computed solutions to some RST routing problems using the answer set solver CMODELS.

Character-Based cladistics and answer set programming

Wire routing is the problem of determining the physical locations of all the wires interconnecting the circuit components on a chip. Since the wires cannot intersect with each other, they are competing for limited spaces, thus making routing a difficult combinatorial optimization problem. We present a new approach to wire routing that uses action languages and satisfiability planning. Its idea is to think of each path as the trajectory of a robot, and to understand a routing problem as the problem of planning the actions of several robots whose paths are required to be disjoint. The new method differs from the algorithms implemented in the existing routing systems in that it always correctly determines whether a given problem is solvable, and it produces a solution whenever one exists.