This project is led by Fredrik Heintz and funded by CENIIT, the Center for Industrial Information Technology.
Table of contents:
An increasing number of products and devices, ranging from mobile phones to trucks and aircraft, are becoming equipped with advanced sensors providing information about the system itself and its environment. In many cases, this wealth of information is not very useful in its raw form. On the other hand, given suitable processing using high-level relational, spatial and temporal reasoning techniques, a great deal of interesting knowledge could be extracted, significantly improving the situational awareness of the system itself and the decision support provided to its human users, something that would greatly improve the usefulness of unmanned aerial vehicles and the fault diagnosis and isolation capability in trucks.
Though a rich and varied set of useful high-level reasoning techniques already exist, they usually require information at a level of abstraction that is far higher than that which is normally acquired through sensing. For example, existing methods for detecting complex events occurring in the environment tend to assume a set of primitive but meaningful symbolic events as input, whereas a camera only provides a raw stream of bitmap images. Thus, there is a wide gap between assumed and provided abstraction levels, which we call the sense-reasoning gap.
Bridging this gap is a fundamental and very challenging problem. Information must continuously be extracted from the environment through all available sensors. Suitable representations of this information must be constructed at a low level. Then, the available information must continuously be processed using well-defined and proven methods, in order to raise its level of abstraction. Since the gap in abstraction levels can be very wide, an incremental methodology is often required, where information passes through a sequence or network of methods that eventually produce the desired knowledge. At all times, information at any level must be properly connected and correlated to information at lower levels, ensuring that all levels of reasoning are grounded in the external environment.
How to bridge the sense-reasoning gap in a principled and systematic manner is largely an unstudied problem in itself. Our hypothesis is that a stream-based approach provides a highly appropriate structure, since streams capture the incremental nature of the information available from sensors and the continuous reasoning process of making inferences with minimal latency necessary to react to rapid changes in the environment.
Stream processing is almost as old as computer science itself [r17], and has been studied quite intensively for the last 10 years. For example, stream data management systems such as STREAM [r18], XStream [r7], Aurora [r2], and Borealis [r2] provide continuous query languages supporting filters, maps, and joins on streams. Similarly, systems for complex event processing, such as Cayuga [r1] and SASE [r8], provide specification languages for defining complex events in terms of more primitive events occurring in event streams.
However, these systems do not provide the appropriate abstractions for bridging the sensereasoning gap by themselves. We have therefore constructed a stream-based framework called DyKnow [r10] which has been used to bridge the gap between sensing and reasoning to a certain degree in sophisticated unmanned aerial vehicle (UAV) applications such as traffic monitoring [r14] and assisting emergency services [r6].
DyKnow provides both conceptual and practical support for stream-based reasoning grounded through sensing by organizing the many levels of information processing in an autonomous system as a coherent network of sensing and reasoning processes connected by streams. DyKnow is fully implemented and a central component in our distributed UAV architecture [r5].
The purpose of this project is to extend the theoretical basis for DyKnow, to develop techniques and algorithms for stream-based reasoning grounded through sensing, and to explore their use in academic and industrial applications in cooperation with companies such as Scania and Saab. The long term vision is to establish a strong cognitive robotics group focusing on formal reasoning techniques based on grounded representations and their use in autonomous systems. DyKnow will be extended to support two concrete application areas Cooperative situational awareness and Reconfigurable diagnosis for large-scale complex systems. By developing the same framework in two rather different directions we make sure that DyKnow provides support for a wide variety of interesting and relevant applications.
As UAV applications are becoming more complex and covering larger physical areas there is an increasing need for multiple UAVs to cooperatively solve problems which are beyond the capability of an individual UAV. One class of such applications is multi-UAV surveillance missions, where a team of UAVs is keeping track of important objects and events in order to create and maintain situational awareness about an area.
Our goal is to create a framework for such cooperative situational awareness applications. We believe that DyKnow provides an appropriate basis for such a framework by supporting sharing and fusing of information from distributed sources on many levels of abstraction. We have previously shown [r11] how DyKnow can be used to implement the JDL Data Fusion Model [r19], which is the de facto standard functional fusion model. This shows that DyKnow has the basic functionality to support fusion applications, even though the support can be further extended.
As a step towards the goal, we have started to develop a DyKnow federation framework for integrating information among multiple agents such as UAVs [r12]. To create and maintain the federation, a multi-agent delegation framework [r4] is used which allows high-level specification and reasoning about resource bounded cooperative problem solving. When the federation is set up, local information is transparently shared between the agents according to specification.
One fundamental problem in situational awareness is how to associate persistent identities with entities found in sensor data. For example, how should a UAV determine whether the blue car it detects now is the same car it saw 20 minutes earlier?
The process of creating and continuously maintaining an association between symbolic identities labeling physical objects and sensor data being collected about them is called anchoring [r3] and is a special case of the general symbol grounding problem [r9]. Cooperative anchoring is the problem of how to anchor symbols to sensor data when information from multiple UAVs is needed to determine the correct identity. For example, for one UAV to determine that the blue car it sees now is indeed the same car it saw before, it may require information about car movements from other UAVs in the area. How to solve this is an open issue, however recent work [r13] provides a promising foundation. If this fundamental problem is solved, then it would be possible to start reasoning about objects over long time periods which is essential for surveillance and other applications where all objects are not seen at all times.
Another interesting question is how a UAV that has anchored a symbol can provide another UAV with enough information for it to anchor the same symbol to its own sensor data. This is the problem of how to share anchors. If the two UAVs have the same sensors and the same processing capabilities then it is quite straight forward to share an anchor, but when they have different sensors then a high-level representation is required which can be anchored internally through the sensors of the other UAV. This is already done to a certain degree in our anchoring approach, since we represent the anchoring of a symbol on many levels of abstraction. Therefore, we believe that our approach can be extended to handle this problem.
The diagnosis problem typically consists of detecting and isolating faulty components given available sensor and actuator signals. Reliable real-time multiple fault diagnosis of dynamical systems in the presence of noise is of fundamental importance in many industrial applications, but is generally associated with a high computational burden.
The FlexDx diagnosis framework is designed to reduce this burden through adaptive reconfiguration, by exploiting the fact that only a subset of all possiblele tests are required in order to detecting that some fault has occurred [7]. In case of an alarm, additional tests can be run, partially on historic sensor information, in order to further isolate the faulty components.
However, special attention has to be given to the issues introduced by reconfigurability. For example, tests are added and removed dynamically, tests are partially performed on historic data, and synchronous and asynchronous processing are combined. Another challenging question is how to select the next set of tests to minimize the time to isolate the fault. How to handle these issues in a principled way will be studied in this track.
To guide the research and to test FlexDx in a realistic application we have applied FlexDx to the Advanced Diagnostics and Prognostics Testbed (ADAPT) defined by NASA Ames and PARC [r15]. It is used in the diagnosis competition at the International Workshop on Principles of Diagnosis (DX) to systematically evaluate different technologies and to produce comparable performance assessments for diagnostics methods. Since ADAPT is a large testbed with many sensors and components that requires more tests than can be ran at the same time, FlexDx is a good starting point.
Another very interesting industrial application of reconfigurable diagnosis, suggested by our industrial partner Scania, is in the health monitoring of a large fleet of heavy trucks. Each of these vehicles has a large number of sensors which continually collect information that is used to monitor its health. However, at construction time it is not known exactly what issues should be monitored and while driving it is not possible to store all the data collected. So, when an issue is reported for one truck it would be highly interesting to be able to automatically reconfigure the diagnosis system on each truck to start monitoring the new issue. In order to achieve this a reconfigurable diagnosis framework is required. Using a DyKnow, a solution of direct industrial relevance for Scania can be developed.
This project takes place within the Artificial Intelligence and Integrated Computer Systems (AIICS) division of IDA. The division has a long-standing track record of excellence in the areas knowledge representation and cooperative multi-agent Unmanned Aircraft Systems (UAS) as well as planning. The group is one of the internationally leading UAS groups in academia and at the front line of research in intelligent autonomous systems. The group has several medium and small scale autonomous UAVs for indoor and outdoor operations. These systems are fully functional and have been used in complex mission scenarios flown autonomously in the Revinge rescue service facility in southern Sweden.
Through the overall UAV project, we cooperate with Saab Aerosystems for example through multiple NFFP projects and through LinkLab, a center for future aviation systems. Results have been reported through both technical reports and peer-reviewed publications, and demonstrated and discussed at several joint workshops and live demonstrations. We also have frequent industrial feedback through Patrick Doherty, who meets Gunnar Holmberg (Saab) regularly to discuss progress within LinkLab. The collaboration has for example resulted in a Master Thesis done at Saab which evaluated DyKnow from their perspective and found it to satisfy their requirements [r16].
We also have active collaboration with Scania, mainly through an industry-based doctoral student (industridoktorand) in our group who works on the use of planning for diagnosis of heavy trucks. This provides an excellent connection to another important industrial partner. His co-adviser docent Mattias Nyberg at Scania is our primary contact person. One purpose of this project is to extend the existing cooperation to the use of stream-based reasoning techniques in Scania's heavy trucks.
AIICS is a member of the VR-funded Linnaeus research environment CADICS and the ELLIIT excellence center funded by the Department of Education's support for strategic research areas. Both centers offers state-of-the-art competence in many multi-disciplinary areas pertinent to this project as well as feedback from their industry connections. Through these centers, Fredrik Heintz has successful collaboration with Mattias Krysander and Erik Frisk from the diagnosis group at ISY which has resulted in the FlexDx diagnosis framework. The diagnosis group also has relevant industrial partners such as Siemens Industrial Turbomachinery and Uptime.
The combination of multiple heterogeneous UAV platforms, strong inter-disciplinary collaboration with industry and academia, and proven research excellence makes this research environment highly suitable for this CENIIT project.
At the moment, the following people are involved in this CENIIT project:
People previously involved in this CENIIT project:
Progress towards our goal of creating a framework for cooperative situational awareness applications has been made along two lines. The main part has consisted of developing a pragmatic framework for collaborative robotics [3] which allows high-level specification and reasoning about resource bounded cooperative problem solving. It it based on previous work at AIICS on a theory of delegation [r4]. There has also been some progress on extending DyKnow for selective sharing and merging of information in a DyKnow Federation to support distributed information fusion among collaborative UAVs [5].
We have also published three overview papers about the use of stream-based reasoning in autonomous systems [1, 2, 4]. Special focus was placed on the support by stream-based reasoning for anchoring and for the pragmatic use of planning.
Work on extending the FlexDx diagnosis framework towards handling the ADAPT test bed has proceeded and has raised a number of interesting questions related to diagnosis. These problems are still being addressed. The work has resulted in a number of publications in the area of diagnosis by the collaboration partners based on our common discussions [r20, r21].
Together with Mattias Nyberg from Scania AB a Master's thesis project "Stream Processing for Vehicle Health Monitoring" has been defined. The purpose is to explore the possibilities of processing streams of data produced by on-board sensors for monitoring the health of a truck. In a first iteration we are interested in collecting information which can be analyzed off-line. The information could for example be in the form of events, partial snapshots of the system state, streams of continuous data, and statistics over streams of data. To specify which information should be collected a specification language is required that can be efficiently implemented on the on-board hardware. An important and interesting consideration is the trade-off between CPU usage, memory consumption, and information quality. For example, when all required information can not be collected how should the system decide what should be approximated or discarded.
The main part of the project has been the continued development of a pragmatic framework for collaborative robotics which allows high-level specification and reasoning about resource bounded cooperative problem solving. An important part of this framework is the use of distributed constraint reasoning for task and resource allocation among teams of collaborative robots. The work has resulted in a number of publications [8, 9, 10, 11, 15, 16] including a book chapter [11] and a licentiate thesis by David Landén co-supervised by Fredrik Heintz [15].
We have also made progress towards providing support for stream-based reasoning in the Robot Operating System (ROS) which is a very popular framework for developing robotic applications. When finished the software will be available as open source. This will allow stream-based reasoning to be used by a wide range of robots including our LinkQuad micro air vehicles and commercially available platforms such as the PR2 from Willow Garage. This work is done in collaboration with Anders Hongslo, a Master's student in the final stages of his thesis. Zlatan Dragisic, another Master's student, has extended DyKnow with a semantic-matching capability which allows symbols in a logical formula to automatically be matched to streams providing an interpretation of the symbol through reasoning about the meaning of the symbols and the content of streams as represented by one or more ontologies [14]. Daniel Lazarovski, a third Master's student, is working on extending the temporal reasoning currently provided by DyKnow with support for spatial reasoning in RCC-8. All three Master's theses are done in close collaboration with me.
The main result is the Master's thesis "Design Patterns for Service-Based Fault Tolerant Mechatronic Systems" by Erik Lundqvist which was done at Scania AB in Södertälje [13]. The thesis studies the use of service-based fault tolerant control on a real system for selective catalytic reduction developed at Scania. It identifies a number of design patterns that are used for typical signal flow architectures in mechatronic systems and extends these design patterns to support fault tolerance according to the service-based fault tolerant control approach. The services in this framework are similar to the knowledge processes in DyKnow and both use streams to communicate (they are called signals in Scania's framework) so the design patterns are relevant for stream-based applications as well.
Another result is a Bachelor's thesis comparing DyKnow to synchronous languages by implementing the FlexDx diagnosis framework in SIGNAL and arguing that DyKnow provides more appropriate support [12].
The main part of the project has been the continued development of a pragmatic framework for collaborative robotics which allows high-level specification and reasoning about resource bounded cooperative problem solving. A central part of this framework is allocating complex tasks to heterogeneous robotic systems. The problem of complex task allocation can be formulated as a distributed constraint satisfaction or optimization problem [15, 11]. Existing algorithms for solving distributed constraint problems do not scale since they do not explicitly consider this class of problems which has conditional constraints. The focus is therefore on developing new algorithms for solving these distributed conditional constraint satisfaction and optimization problems. These algorithms will be of general use and could be interesting to our industrial partners.
We have also made progress towards providing support for stream-based reasoning in the Robot Operating System (ROS). Anders Hongso has completed his Master's thesis on stream reasoning in ROS [20]. Daniel Lazarovski has completed his Master's thesis on extending the temporal reasoning provided by DyKnow with support for spatial reasoning in RCC-8 [17]. This allows spatio-temporal reasoning over streaming data in DyKnow which is an important extension. A paper based on the Master's thesis work of Zlatan Dragisic [14] on semantic information integration for stream reasoning was presented at the international Fusion Conference [21]. Recently we have submitted a journal paper on stream-based hierarchical anchoring to a special issue on Symbol Grounding.
Together with Unmesh Bordoloi I have supervised a Master's thesis about formal timing analysis of stream reasoning [19]. The idea is to use formal analysis tools from real time systems to do design space exploration of stream reasoning applications. Using these tools it is possible for example to analyze how many temporal logical formulas can be evaluated given a particular sampling frequency. It is also possible to analyze the effect on delays in the stream reasoning based on the jitter in the delivery of the different streams. This is a cooperation with Bordoloi's CENIIT project A Cross-layer Approach to Reliability Optimization for Automotive Electronic Systems.
Together with the Daniel de Leng very good progress has been made on both the pragmatic and theoretical aspects of grounded stream-based reasoning. An important step forward is the completion of a first version of DyKnow for the Robot Operating System (ROS) [25]. This provides support for grounded stream-based reasoning for a wide range of robotic applications. A second result is extending the semantic grounding functionality of DyKnow with support for finding and automatically applying transformations [24]. This allows situations when the requested information is not directly available but has to be generated or adapted through transformations to be handled. Two types of transformations are considered. Automatic transformation between different units of measurements and between streams of different types.
The journal paper on stream-based hierarchical anchoring submitted in 2012 was accepted and published [22].
A journal paper describing the latest version of our pragmatic collaborative robotics framework was completed and published [23].
Another important result is that Fredrik Heintz submitted his docent application in September 2013.
Three very important events for 2014 are that Fredrik Heintz is now a docent in computer science at Linköping University, that Daniel de Leng finished his Master's Thesis [27] and was awarded the Swedish AI Society's award for best AI Master's Thesis, and that Daniel de Leng is now employed as a PhD student funded by a new CUGS grant and this CENIIT grant.
Together with the Daniel de Leng very good progress has been made on both the pragmatic and theoretical aspects of grounded stream-based reasoning. We have continued the work on extending the temporal reasoning provided by DyKnow with support for spatial reasoning in for example RCC-8 [30]. This allows for efficient spatio-temporal reasoning over streaming data. The work on semantic matching has also been further extended, this time to allow initial integration with semantic event processing [28]. We have also continued to improve the Robot Operating System (ROS) version of DyKnow.
Together with Mattias Tiger, a very good Master's Student, we have studied how unsupervised learning can be used to automatically learn activities and how these activities relate to each other based on observations of trajectories of objects. The idea is to take a bottom-up approach to learning the hierarchical spatio-temporal patterns that our stream reasoning framework can then detect and reason about. The initial results have been presented at the European Starting AI Researcher Symposium [29].
The following publications have been published or accepted as of September 14, 2014.
| [30] | Fredrik Heintz and Daniel de Leng. Spatio-Temporal Stream Reasoning with Incomplete Spatial Information. In Proc. European Conference on Artificial Intelligence (ECAI), 2014. [ PDF ] |
| [29] | Mattias Tiger and Fredrik Heintz. Towards Learning and Classifying Spatio-Temporal Activities in a Stream Processing Framework. In Proc. European Starting AI Researcher Symposium (STAIRS), 2014. [ PDF ] |
| [28] | Daniel de Leng and Fredrik Heintz. Towards On-Demand Semantic Event Processing for Stream Reasoning. In Proc. International Conference on Information Fusion (FUSION), 2014. [ PDF ] |
| [27] | Daniel de Leng. Extending semantic matching in DyKnow to handle indirectly-available streams. Master's Thesis, Utrecht University, 2013. [ PDF ] |
| [26] | Patrick Doherty and Fredrik Heintz and Jonas Kvarnström. Robotics, Temporal Logic and Stream Reasoning. In Proc. Logic for Programming Artificial Intelligence and Reasoning (LPAR), 2013. [ PDF ] |
| [25] | Fredrik Heintz. Semantically Grounded Stream Reasoning Integrated with ROS. In Proc. IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2013. [ PDF ] |
| [24] | Fredrik Heintz and Daniel de Leng. Semantic Information Integration with Transformations for Stream Reasoning. In Proc. International Conference on Information Fusion (FUSION), 2013. [ PDF ] |
| [23] | Patrick Doherty and Fredrik Heintz and Jonas Kvarnström. High-level Mission Specification and Planning for Collaborative Unmanned Aircraft Systems using Delegation. Journal of Unmanned Systems 1(1):75-119, World Scientific, 2013. [ PDF ] |
| [22] | Fredrik Heintz and Jonas Kvarnström and Patrick Doherty. Stream-Based Hierarchical Anchoring. Künstliche Intelligenz 27(2):119-128, Springer, 2013. [ PDF ] |
| [21] | Fredrik Heintz and Zlatan Dragisic. Semantic Information Integration for Stream Reasoning. In Proc. of the International Conference on Information Fusion (FUSION), 2012. [ PDF ] |
| [20] | Anders Hongslo. Stream Processing in the Robot Operating System framework. Master's Thesis, Linköping University, 2012. [ Diva ] |
| [19] | Viet Ha Nguyen. Design Space Exploration of the Quality of Service for Stream Reasoning Applications. Master's Thesis, Linköping University, 2012. [ Diva ] |
| [18] | Patrick Doherty and Fredrik Heintz. Delegation-Based Collaboration. In Proc. International Conference on Cognitive Systems (CogSys), 2012. |
| [17] | Daniel Lazarovski. Extending the Stream Reasoning in DyKnow with Spatial Reasoning in RCC-8. Master's Thesis, Linköping University, 2012. [ Diva ] |
| [16] | Patrick Doherty and Fredrik Heintz. A Delegation-Based Cooperative Robotic Framework. In Proc. IEEE International Conference on Robotics and Biomimetics (ROBIO), 2011. [ PDF ] |
| [15] | David Landén. Complex Task Allocation for Delegation: From Theory to Practice. Licenciate Thesis, Linköping University, 2011. [ Diva ] |
| [14] | Zlatan Dragisic. Semantic Matching for Stream Reasoning. Master's Thesis, Linköping University, 2011. [ Diva ] |
| [13] | Erik Lundqvist. Design Patterns for Service-Based Fault Tolerant Mechatronic Systems. Master's Thesis, Linköping University, 2011. [ Diva ] |
| [12] | Anders Skoglund. Comparing Asynchronous and Synchronous Approaches to Knowledge Processing. Bachelor's Thesis, Linköping University, 2011. [ Diva ] |
| [11] | Patrick Doherty, Fredrik Heintz, and David Landén. A Delegation-Based Architecture for Collaborative Robotics. In D. Weyns and M.-P. Gleizes (Eds.): AOSE 2010, LNCS 6788, pp. 205-247, Springer-Verlag, 2011. [ PDF ] |
| [10] | Patrick Doherty, Fredrik Heintz and David Landén. A Delegation-Based Collaborative Robotic Framework. In Proc. International Workshop on Collaborative Agents - Research and development (CARE), 2011. [ PDF ] |
| [9] | David Landén, Fredrik Heintz and Patrick Doherty. Complex Task Allocation in Mixed-Initiative Delegation: A UAV Case Study (Early Innovation). In Proc. International Conference on Principles and Practice of Multi-Agent Systems (PRIMA), 2010. [ PDF ] |
| [8] | Patrick Doherty, David Landén and Fredrik Heintz. A Distributed Task Specification Language for Mixed-Initiative Delegation. In Proc. International Conference on Principles and Practice of Multi-Agent Systems (PRIMA), 2010. [ PDF ] |
| [7] | Mattias Krysander, Fredrik Heintz, Jacob Roll and Erik Frisk. FlexDx: A reconfigurable diagnosis framework. Engineering applications of artificial intelligence, 23(8):1303-1313, 2010. [ PDF ] |
| [6] | Fredrik Heintz, Jonas Kvarnström and Patrick Doherty. Bridging the sense-reasoning gap: DyKnow - Stream-based middleware for knowledge processing. Advanced Engineering Informatics 24(1):14-26, 2010. [ PDF ] |
| [5] | Fredrik Heintz and Patrick Doherty. Federated DyKnow, a Distributed Information Fusion System for Collaborative UAVs. In Proc. International Conference on Control, Automation, Robotics and Vision (ICARCV), 2010. [ PDF ] |
| [4] | Fredrik Heintz, Jonas Kvarnström and Patrick Doherty. Stream-Based Reasoning Support for Autonomous Systems. In Proc. European Conference on Artificial Intelligence (ECAI), 2010. [ PDF ] |
| [3] | Patrick Doherty, Jonas Kvarnström, Fredrik Heintz, David Landén, and Per-Magnus Olsson. Research with Collaborative Unammnned Aircraft Systems. In Proc. Dagstuhl Workshop on Cognitive Robotics, 2010. |
| [2] | Fredrik Heintz, Jonas Kvarnström and Patrick Doherty. Stream-Based Reasoning in DyKnow. In Proc. Dagstuhl Workshop on Cognitive Robotics, 2010. |
| [1] | Fredrik Heintz, Jonas Kvarnström and Patrick Doherty. Stream-Based Middleware Support for Embedded Reasoning. In Proc. AAAI Spring Symposium Embedded Reasoning: Intelligence in Embedded Systems, 2010. [ PDF ] |
| [r1] | Lars Brenna, Alan Demers, Johannes Gehrke, Mingsheng Hong, Joel Ossher, Biswanath Panda, Mirek Riedewald, Mohit Thatte, and Walker White. Cayuga: a high-performance event processing engine. In Proc. SIGMOD, pages 1100-1102, 2007. [ http ] |
| [r2] | U. Çetintemel, D. Abadi, Y. Ahmad, H. Balakrishnan, M. Balazinska, M. Cherniack, J. Hwang, W. Lindner, S. Madden, A. Maskey, A. Rasin, E. Ryvkina, M. Stonebraker, N. Tatbul, Y. Xing, and S. Zdonik. The Aurora and Borealis Stream Processing Engines. In Data Stream Management: Processing High-Speed Data Streams. 2007. |
| [r3] | S. Coradeschi and A. Saffiotti. An introduction to the anchoring problem. Robotics and Autonomous Systems, 43(2-3):85-96, 2003. |
| [r4] | P. Doherty and J-J. Ch. Meyer. Towards a delegation framework for aerial robotic mission scenarios. In Proc. CIA, pages 5-26, 2007. |
| [r5] | Patrick Doherty, Patrik Haslum, Fredrik Heintz, Torsten Merz, Per Nyblom, Tommy Persson, and Björn Wingman. A distributed architecture for autonomous unmanned aerial vehicle experimentation. In Proceedings of the 7th International Symposium on Distributed Autonomous Robotic Systems (DARS), pages 221-230, Toulouse, France, June 2004. [ Conference | .pdf ] |
| [r6] | Patrick Doherty, Jonas Kvarnström, and Fredrik Heintz. A temporal logic-based planning and execution monitoring framework for unmanned aircraft systems. Journal of Autonomous Agents and Multi-Agent Systems, 19(3):332-337, October 2009. First published online February 2009. [ Journal | .pdf ] |
| [r7] | L. Girod, Yuan Mei, R. Newton, S. Rost, A. Thiagarajan, H. Balakrishnan, and S. Madden. Xstream: a signal-oriented data stream management system. In Proc. ICDE, pages 1180-1189, 2008. [ http ] |
| [r8] | Daniel Gyllstrom, Eugene Wu, Hee-Jin Chae, Yanlei Diao, Patrick Stahlberg, and Gordon Anderson. Sase: Complex event processing over streams. In Proc. CIDR, 2007. |
| [r9] | S. Harnad. The symbol-grounding problem. Physica, D(42):335-346, 1990. |
| [r10] | Fredrik Heintz. DyKnow: A Stream-Based Knowledge Processing Middleware Framework. Ph.d. thesis, Linköping University, Department of Computer and Information Science, March 2009. [ .pdf | E-Press ] |
| [r11] | Fredrik Heintz and Patrick Doherty. A knowledge processing middleware framework and its relation to the jdl data fusion model. Journal of Intelligent and Fuzzy Systems, 17(4):335-351, February 2006. [ Journal | .pdf ] |
| [r12] | Fredrik Heintz and Patrick Doherty. DyKnow federations: Distributing and merging information among UAVs. In Proceedings of the 11th International Conference on Information Fusion (Fusion), Cologne, Germany, July 2008. [ Conference | .pdf ] |
| [r13] | Fredrik Heintz, Jonas Kvarnström, and Patrick Doherty. A stream-based hierarchical anchoring framework. In Proceedings of the IEEE/RSJ International Conference on Intelligent RObots and Systems (IROS), St. Louis, Missouri, October 2009. [ Conference ] |
| [r14] | Fredrik Heintz, Piotr Rudol, and Patrick Doherty. From images to traffic behavior - a UAV tracking and monitoring application. In Proceedings of the 10th International Conference on Information Fusion (Fusion), Quebec, Canada, July 2007. ISIF, IEEE, AES. [ Conference | .pdf ] |
| [r15] | T. Kurtoglu, S. Narasimhan, S. Poll, D. Garcia, L. Kuhn, J. de Kleer, A. van Gemund, and A. Feldman. Towards a framework for evaluating and comparing diagnosis algorithms. In Proc. DX, 2009. |
| [r16] | Tommy Persson. Evaluating the use of DyKnow in multi-UAV traffic monitoring applications. Master's thesis, Linköpings universitet, 2009. |
| [r17] | R. Stephens. A survey of stream processing. Acta Informatica, 34(7):491-541, 1997. |
| [r18] | The STREAM Group. STREAM: The Stanford stream data manager. IEEE Data Engineering Bulletin, 26(1), 2003. [ http ] |
| [r19] | F. White. A model for data fusion. In Proc. National Symposium for Sensor Fusion, 1988. |
| [r20] | E. Frisk, A. Bregon, J. Åslund, M. Krysander, B. Pulido, and G. Biswas. Diagnosability Analysis Considering Causal Interpretations for Differential Constraints In Proc. DX, 2010. |
| [r21] | M. Krysander, J. Åslund and E. Frisk. A Structural Algorithm for Finding Testable Sub-models and Multiple Fault Isolability Analysis. In Proc. DX, 2010. |