Swedish Defence Research Agency

About: Swedish Defence Research Agency is a government organization based out in Stockholm, Sweden. It is known for research contribution in the topics: Radar & Laser. The organization has 1413 authors who have published 2731 publications receiving 56083 citations. The organization is also known as: Totalförsvarets forskningsinstitut.

Capabilities-based plan recognition

Abstract: The new types of opponents and new kinds of situations that the Swedish defence forces are facing today calls for new information fusion methods. In order to provide commanders with the ability to predict the enemy's future actions, tools for automatic plan recognition are needed. In this paper, we take the first step towards constructing such a method based on recognizing plans using information about the capabilities of the enemy. The method combines our previous work on plan recognition using bayesian networks based on comparing enemy movements to their doctrines and methodology for force aggregation using capabilities. We describe how the plans of the enemy are built up so that their intended effects are achieved. The relations between these, their resources and the context in which they are acting are used to construct the plan recognition network. We discuss the need for including termination states in the plan recognition method and describe ontologies that are aimed to support the construction of the bayesian networks needed for capability-based plan recognition. We conclude with a discussion of possible extensions of the method.

Recognition of Anomalous Motion Patterns in Urban Surveillance

Abstract: We investigate the unsupervised K-means clustering and the semi-supervised hidden Markov model (HMM) to automatically detect anomalous motion patterns in groups of people (crowds). Anomalous motion patterns are typically people merging into a dense group, followed by disturbances or threatening situations within the group. The application of K-means clustering and HMM are illustrated with datasets from four surveillance scenarios. The results indicate that by investigating the group of people in a systematic way with different K values, analyze cluster density, cluster quality and changes in cluster shape we can automatically detect anomalous motion patterns. The results correspond well with the events in the datasets. The results also indicate that very accurate detections of the people in the dense group would not be necessary. The clustering and HMM results will be very much the same also with some increased uncertainty in the detections.