Flexible solutions in the field of intralogistics are needed more and more because of a higher dynamic in the environment. A possible solution to achieve higher flexible and are robuster system is the use of an automated guided vehicle (agv) system. In this paper current research in the field of decentralized agv system control i.e. in conflict free routing and in decentralized order allocation is described. Furthermore approaches to improve self-control by introducing negotiations are presented as well as the challenges that arise when such approaches should be realized.
Shahin Shah, Lukáš Chrpa, Peter Gregory, Thomas. L. Mccluskey, Falilat Jimoh
282 - 293
An important area in AI Planning is the expressiveness of planning domain specification languages such as PDDL, and their aptitude for modelling real applications. This paper presents OCLplus, an extension of a hierarchical object centred planning domain definition language, intended to support the representation of domains with continuous change. The main extension in OCLplus provides the capability of interconnection between the planners and the changes that are caused by other objects of the world. To this extent, the concept of event and process are introduced in the Hierarchical Task Network (HTN), object centred planning framework in which a process is responsible for either continuous or discrete changes, and an event is triggered if its precondition is met. We evaluate the use of OCLplus and compare it with a similar language, PDDL+.
We address the problem of mining data streams using Artificial Neural Networks (ANN). Usual data stream clustering models (eg. k-means) are too dependent on assumptions regarding cluster statistical properties (ie. number of clusters, cluster shape), while unsupervised ANN algorithms (Adaptive Resonant Theory — ART networks and Self-Organizing Maps — SOM) are recognized widely by their ability to discover hidden patterns, generalization capabilities and robustness to noise. However, use of ANNs with the data stream model is still poorly explored. We propose a methodology and modular framework to cluster data streams and extract other relevant knowledge. Empirical results with both synthetic and real data provide evidence of the validity of the approach.
This paper explores how current planners behave when exposed to multiple metrics, examining which of the planners are metric sensitive and which are not. For the metric insensitive planners we propose a new method of simulating metric sensitivity for the purpose of generation of diverse plans close to a pareto frontier. It is shown that metric sensitive planners are good candidates for generating sets of pareto optimal plans.
Spatial assistance systems designed to empower people in smart environments need to perceive their operational environment, recognize activities performed in the environment, and reason about the observed information in order to plan a course of action. Activities performed by humans are spatio-temporal interactions between a subject, objects, and the (potential) group-based context in which they are performed. Activities mediate and develop space and manifest in spatio-temporal interactions of humans with the environment and the artefacts within. We propose a human-centred activity-theoretic model for the description of activities by their motives and goals. The model itself is grounded with respect to the spatio-temporal interactional characteristics of the activities being modelled. This description serves as a first step towards bridging the gap between sensor readings and high-level reasoning about space, actions, and change within a logic-based commonsense reasoning framework. To illustrate our ideas we introduce a work-in-progress smart meeting scenario, an overarching scenario that provides a developmental basis for the ongoing doctoral research described in this contribution.
Teo Susnjak, Andre Barczak, Napoleon Reyes, Ken Hawick
330 - 335
We propose a general method applicable to existing multiclass boosting-algorithms for creating cascaded classifiers. The motivation is to introduce more tractability to machine learning tasks which require large datasets and involve complex decision boundaries, by way of separate-and-conquer strategies that reduce both the training and detection-phase overheads. The preliminary study explored the application of our method to AdaBoost.ECC on six UCI datasets and found that a decrease in the computational training and evaluation overheads occurred without significant effects on the generalization of the classifiers.
In this paper, we revisit the idea of splitting a planning problem into subproblems hopefully easier to solve with the help of landmark analysis. This technique initially proposed in the first approaches related to landmarks in classical planning has been outperformed by landmark-based heuristics and has not been paid much attention over the last years. We believe that it is still a promising research direction, particularly for devising distributed search algorithms that could explore different landmark orderings in parallel. To this end, we propose a new method for problem splitting based on landmarks, which has three advantages over the original technique: it is complete (if a solution exists, the algorithm finds it), it uses the precedence relations over the landmarks in a more flexible way (the orderings are explored by way of a best-first search algorithm), and finally it can be easily performed in parallel (by e.g. following the hash-based distribution principle). We lay in this paper the foundations of a meta best-first search algorithm, which explores the landmark orderings and can use any embedded planner to solve each subproblem. It opens up avenues for future research: among them are new heuristics for guiding the meta search towards the most promising orderings, different policies for expanding nodes of the meta search, influence of the embedded subplanner, and parallelization strategies of the meta search.
In this paper, we propose a normative Multi-Agent System to handle uncertainty in a monitoring application. It is based on the assertion that no single most-likely situation should be considered, thus requiring the management of multiple concurrent hypotheses. A decision is then made by comparing these hypothesized situations to requirements and expectations, thus detecting potential problems. This system uses a large knowledge base of interconnected situation models on several levels of abstraction. It is centered around the need to constantly reconsider which hypotheses should be evaluated, with regards to both the current data from the sensors and wider requirements in terms of efficiency and specific focus from an expected scenario. We propose both a generic concept, and a more specific system for human health monitoring, using ambulatory physiological sensors.