Video Camouflaged Object Detection (VCOD) aims to segment objects visually indistinguishable from their background in unconstrained video sequences. This task is challenging due to the extremely low appearance contrast and motion ambiguity caused by camera shifts, occlusion, and background clutter. Although motion cues provide crucial signals to break camouflage, existing methods typically adopt either implicit motion modeling based on initial semantic representations or explicit modeling based on low-level pixel motion, both of which suffer from information loss or noise sensitivity. To this end, we propose IDM-VCOD, an integrated dual-motion VCOD framework that unifies two complementary motion modeling strategies through a dual-motion integrated design. The implicit path aggregates temporal context via spatio-temporal prediction cubes for temporal neighborhood refinement, while the explicit path explicitly aligns multi-frame backgrounds to highlight foreground motion. A selective activation mechanism adaptively triggers the explicit branch only when implicit predictions are unreliable, enhancing accuracy and efficiency. Extensive experiments on the MoCA-Mask and CAD datasets demonstrate that IDM-VCOD achieves superior detection accuracy and generalization compared to state-of-the-art VCOD methods, while significantly reducing model size and inference cost. Comprehensive ablation studies further validate the effectiveness of the dual-motion design and its robustness in challenging camouflage scenarios.

The ongoing evolution of Software Defined Vehicles (SDVs) in the automotive industry has drawn attention to the importance of decoupling hardware and software to enable greater flexibility, upgradability, and adaptability in vehicle design and functionality. Digital twin technology, which involves creating virtual models of physical systems with bidirectional communication with the physical system, presents a promising approach for optimizing and validating various aspects of SDVs within this new paradigm. This paper proposes a novel framework for applying digital twins in the context of SDVs, and the benefits derived from decoupling hardware and software.

Rapid advances in artificial intelligence (AI) in the last decade have been largely built upon the wide applications of deep learning (DL). However, the high carbon footprint yielded by larger and larger DL networks has become a concern for sustainability . Furthermore, DL decision mechanism is somewhat obscure in that it can only be verified by test data. Green learning (GL) is being proposed as an alternative paradigm to address these concerns. GL is characterized by low carbon footprints, lightweight model, low computational complexity , and logical transparency. It offers energy-efficient solutions in cloud centers as well as mobile/edge devices. GL also provides a more transparent, logical decision-making process which is essential to gaining people’s trust. Several statistical tools such as unsupervised representation learning , supervised feature learning, and supervised decision learning, have been developed to achieve this goal in recent years. We have seen a few successful GL examples with performance comparable with state-of-the-art DL solutions. This paper introduces the key characteristics of GL, its demonstrated applications, and future outlook.

Developing models of complex systems, such as Autonomous Vehicles (AVs) that operate in uncertain, reactive environments is challenging because of complexity and uncertainty in their operational environments. Partially Observable Markov Decision Process (POMDP) is a mathematically principled framework that is suitable for developing such models. A POMDP model is typically defined with respect to a specific task performed in the environment and the resulting policy is obtained based on the defined task and information available about the environment. This implies that if changes occur in the task and/or environment, the model and policy need to be updated. One efficient way to achieve this end is to reuse the experience from the initial task(s) to perform similar task(s) instead of designing new models from scratch. This technique is referred to as Transfer Learning (TL). In this paper, we propose a novel TL technique that uses a customized Q-learning algorithm for policy transfer in POMDPs. We use this technique to transfer policies from a POMDP model, initially designed for a simple AV lane-keeping task within a freeway environment, to a similar task in a more complex environment, where random and risky behaviors from neighboring vehicles in the environment can be expected.

Model-Based Systems Engineering (MBSE) requires more significant investment than traditional systems engineering in the early system life cycle phases. Program management justifies this additional investment by arguing that such investments can be expected to produce continuous gains across later phases of the systems life cycle resulting from early detection of defects, risk reduction, improved communication, superior supply chain integration, product line definition, and enhanced traceability. Since systems continue to evolve over their life cycle, system models need to be updated continually to reflect the system's current state and thereby sustain value. However, in today's systems engineering organizations, the current practice is to reallocate modeling resources to other projects once initial modeling on a particular project is completed. This practice results in a resource shortfall that impedes the ability to continuously update system models through the later phases of the system life cycle. This paper presents how digital twin technology can be exploited within MBSE to ensure continuous model updates throughout the system life cycle. This paper also presents preliminary results from prototyping and experiments conducted with a digital twin that is continuously updated with data from the physical system operating in the real world. Finally, this paper shows how operational analysis and system modeling can be significantly enhanced by leveraging digital twin technology within MBSE.

As Model Based Systems Engineering (MBSE) continues to advance in terms of system life cycle coverage, modeling languages, methods, and tools, there is a growing need of an overarching framework for organizing MBSE artifacts that facilitates their rapid retrieval and use by MBSE researchers. At the same time, researchers must have an environment supportive of exploring, experimenting with, and collecting performance data when using potentially heterogenous modeling constructs and algorithms over broad ranges of conditions and assumptions. These requirements jointly imply the need for a MBSE research testbed that enables experimentation with diverse modeling, analysis, simulation, verification, and validation approaches under nominal and off-nominal conditions, collect and analyze data to uncover patterns and trends, reuse models and components as applicable, and serve as a repository for scenarios, models, case studies, and lessons learned. This paper presents progress to date and lessons learned from the prototype MBSE testbed implementation.

Partially Observable Markov Decision Process (POMDP) models are a popular probabilistic modeling method for continuous planning in systems that operate in partially observable, uncertain environments. This paper presents an online algorithm, N-Step Look-Ahead, for solving POMDP models with specific characteristics: flexible state-space, dynamic model parameters (e.g. probability distributions), and real-time constraints (e.g. response needed in fractions of seconds). This algorithm computes the best executable policy by creating a belief-tree starting from the current belief state and employing a look-ahead search over a finite time horizon. To address time-accuracy trade-offs, various heuristics in combination with a distance-based clustering technique is employed to expand and explore only high value belief nodes. To evaluate the accuracy of our algorithm, we compare the online policies computed from N-Step Look-Ahead with offline policies calculated using a customized Q-learning algorithm. We show that the online algorithm can compute optimal policies for beliefs, where belief probabilities are not normally distributed, by looking only a few steps ahead in the belief tree. We discuss computing online policies for normally distributed belief states using our algorithm and explain why they can be different from that obtained from the offline algorithm. Finally, we show how useful heuristics can be developed from Q-learning results to improve the N-Sep Look-Ahead in terms of computation time and accuracy, especially for large POMDP models.

Reuse of proven designs is a common practice in complex engineering systems. However, contextual differences between missions or products - that is, differences in objectives, destinations, environments, and constraints - mean that design reuse can rarely be a "copy and paste" effort. Some amount of rework can be expected to adapt designs from their native or original missions to new missions of interest. This paper develops a generalized process-driven approach for assessing the required rework effort to adapt a given design for a new mission or system. A more structured approach than what is available in the literature and practice, to date, is presented. The paper also describes how the process may be implemented primarily using SysML within an MBSE environment and thereby leverage the significant benefits of MBSE namely, digitizing and centralizing system information, and facilitating reuse via superior storage and access to information. The resulting process forms part of a larger MBSE Reuse Methodology for supporting system architects in making design reuse decisions.

How architects decompose a high-level system architecture often has an outsized impact on subsystem interface complexity and integration strategy. In practice system decomposition tends to be based on experiential know-how, structure of the performing organization, and legacy constraints. Architectural decomposition needs to take other factors into account such as management culture, political considerations, regulations, and social factors. It is important to ensure that the decomposition modularizes the system to reduce systemic complexity. A “best practice” decomposition today calls for a layered architecture with the layers corresponding to operation, function, form, implementation, and organization. How well a particular decomposition works ultimately determines the effectiveness of system integration strategy and resultant system complexity. At each architectural level, the system exhibits a certain degree of complexity. Therefore, aligning modules at each level with the level above can significantly reduce overall system complexity. In this paper, we propose a methodology based on an Inter-Level and Intra-Level Dependency Matrix (I2DM), a dependency-focused specialization of the N2 matrix. When combined with optimization algorithms, and alignment heuristics, this combination helps uncover problematic interactions that could potentially pose integration challenges down the line, and produce unintended, undesirable outcomes. This paper evaluates genetic algorithm with heuristics for system decomposition. The effectiveness of this algorithm is analyzed and an approach for developing an effective integration strategy is presented.

Consistent and accurate understanding of trust and reputation in multi-agent systems is a prerequisite for evaluating system state and determining any needed corrective actions that preserve continued safe operation. In this paper we build on our prior reputation analysis work, which was based on evaluating satisfaction of transactions between agents and agent health. The evaluation was used to discount untrustworthy inputs to a Markov decision model that determines the actions taken by agents in a resilient system network. We extend our earlier work by including new health factors in the trust estimation along with a mechanism for updating each agent's belief state computation through modification of its emission probabilities.

Complex systems are difficult to analyze because of unknown interactions and dependencies among system components, and between the system and the environment. These interactions and dependencies tend to produce unpredictable behaviors that can often lead to detrimental outcomes. Traditional systems engineering that relies on reductive approaches are not suitable for the analysis and design of complex systems. This recognition has led to the development of new paradigms, methods, and tools that better enable: exploration of a complex system's behavior and tradespace; detection, diagnosis, and visualization of component/subsystem dependencies and interactions; and identification, alerting, and circumvention of potentially undesirable interactions. This paper presents a system model-driven interactive storytelling approach for analyzing interactions and dependencies within complex systems, and between complex systems and the environment. An exemplar complex system is used to convey key elements of the approach. The methodology is applicable for a variety of complex systems such as global supply chains, power and energy management, transportation networks, aerospace and aviation enterprises, and defense systems.

Contract-based design (CBD) employs formalisms that explicitly define system requirements, constraints, and interfaces. This paper explores a contract-based design paradigm for expressing system resiliency features. Specifically, resilience formalisms are defined in terms of invariant and flexible assertions. A flexible assertion is one that is learned during system operation and can accommodate unpredicted system behaviors. Invariant assertions are fixed system constraints that are known a priori. A general model structure comprising four key features that contribute to system resilience is presented. In particular, the concept of flexible contracts is operationalized using the Hidden Markov Model (HMM) construct. A system architecture based on flexible contracts and lightweight error monitoring and resiliency response mechanisms is also presented. The proposed framework can serve as a testbed to experiment with different systems resiliency approaches.

The autonomous behavior of tiny, low powered programmable sensing devices often requires constant monitoring for accurate and reliable functioning and data delivery. The problem of monitoring these sensing devices is generally hard due to the dense deployment. It is common to divide such a large problem (or system) into sub-problems (or subsystems) and solve. In sensor networks, we divide the large deployment of sensors into clusters and monitor these groups of sensors or clusters. In this paper, we propose a novel methodology to analyze the critical parameters (bandwidth usage, network congestion, number of dead nodes, activity in the region, and overall energy (power) consumption) of a given network of sensing devices in order to increase the lifetime and reliability of the network.

Information superiority is a source of potential advantage in modern warfare. However, to realize the potential benefits of information superiority require available data and information to be rapidly transformed into actionable knowledge - the key to decision superiority. Decision superiority requires a thorough understanding of the C/sup 2/ decision making processes and the C/sup 2/ information systems that support them. It is with this understanding that we are developing DecisionEdge/spl trade/, a scenario-driven, simulation-based training system that is intended to enable trainees to learn both C/sup 2/ decision making processes as well as C/sup 2/ information systems that support them. DecisionEdge employs a learning objective-guided simulation (LOGS) approach that engages the learner in a first-person, "learn-by-doing" experience. The LOGS approach allows for progressively more difficult training scenarios through systematic manipulation of event density, time stress, and information availability. This learning strategy allows learners to experience a variety of increasingly challenging operational scenarios at a rate commensurate with their learning pace while continuing to keep them involved and motivated.

The infusion of cognitive engineering (CogE) methods and tools into traditional systems engineering processes and practices is becoming an important priority as the systems engineering community begins to tackle system of systems (SoS) problems. As one illustration, the military is interested in creating cognitively-inspired systems (as small as a PDA and as large as a weapon platform) that maximally exploit human potential while also accelerating both human supported and automated decision making. To date, cognitive engineering has been successfully applied to the planning and requirements definition phases of systems engineering (also known as front-end analysis) but has yet to be infused into the remaining phases of systems engineering. To span the full systems engineering lifecycle, it is important, first and foremost, to communicate the return-on-investment to the various stakeholders to get their buy in. The second question that needs to be answered is whether or not "cognitive engineering is ready for primetime". This paper presents promising technical and management strategies to overcome the challenges in introducing cognitive engineering into system-of-systems engineering (SOSE). Specifically, it presents a representative set of cognitive engineering methods and tools to span the full SoS life-cycle as well as cognitive engineering awareness initiatives to penetrate both the DoD acquisition community and commercial industry.

System of systems engineering (SoSE) extracts value from existing assets and designs new assets to be more easily re-purposed, than has been the case. One way SoSE arrives at a system of systems (SoS) is by interfacing or incorporating existing systems. Another way is by "harmonizing" a set of holons. Either way managing the on-going evolution of the SoS is more challenging than initializing the SoS. This paper presents the challenges and opportunities for the next generation of concepts, principles, methods and tools that are needed for creating and sustaining SoS's.

A solid-state inertial measurement unit (IMU) referred to as the MotionPak/sup TM/ is described. The IMU developed by BEI Sensors & Systems Company, consists of three micromachined quartz rate sensors and three accelerometers to provide a six degree of freedom (DOF), miniature, high reliability, low cost motion sensor which Carnegie Mellon University is utilizing to adapt to field robotic applications. Theory of operation and applications of the IMU are presented together with preliminary results in fusing the IMU data with a compass for a field robotics application.

The design of complex systems is a complex process that requires multiple tradeoffs and design iterations. It involves multiple core competencies that seldom reside in a single organization or geographical site. It typically requires multiple design tools that come from different vendors. Design process management in such an environment is key to infusing discipline as well as reducing design cycle time. This paper presents an innovative CORBA-compliant, Web-based, adaptive design process management approach and software tool that is based on dynamically constructed flows. At the heart of this system is an object-oriented ontology of the design process that captures and inter-relates all key objects that are required to manage and improve the design process. A key aspect of the design process manager is its multi-perspective visualization capability that allows both managers and designers to view the design process from their respective perspectives. This paper presents design process management for the multichip module (MCM) design problem. The approach applies to managing the design of any complex system.

IDEON/sup TM//IPPD is the ontology for modeling, analyzing, and managing processes within the systems engineering enterprise. IDEON/sup TM//IPPD is an extension of IDEON/sup TM/, an enterprise ontology that provides the conceptual foundation for modeling, analyzing, integrating, managing, and aiding the execution of enterprise processes. This paper presents the key issues in creating the ontology, discusses the underlying design principles and functional requirements, and describes the various entity types and their relationships.

In an era of ever-increasing global competition, product development and systems engineering organizations are turning to integrated product-process development (IPPD) and process support technologies to improve cycle times, lower costs, and mitigate risks in system development. IPPD is the concurrent development of a product (i.e., system) along with the processes by which the product is created and supported. This paper describes ProcessIPPD/sup TM/, a methodology and toolkit for process modeling and analysis in support of IPPD-enabled collaborative product development and complex systems engineering. Specifically, this paper presents the core IPPD ontology (i.e., key concepts and relationships), and discusses the applications and benefits of process modeling and analysis in IPPD-enabled systems engineering. A sample case study that illuminates the importance of process modeling and analysis is also presented.

Forming or participating in a virtual enterprise is considered an effective strategy to exploit market opportunities that cannot be pursued individually by the partnering organizations. In practice, however, virtual enterprise formation is quite complicated. There are no systematic methodologies or tools to assist decision-makers in: (a) planning and analyzing the formation of a virtual enterprise; and (b) selecting compatible partners/suppliers with the requisite core competencies. This paper presents ProcessWeb/sup TM/, a Web-based, process-enabled, decision support system that assists decision-makers in evaluating potential partnering and outsourcing options.

A trainable model of the human operator in information acquisition tasks is described. The purpose of this model, called the adaptive information selector (AIS), is to select and present textual messages automatically to users in computer-based tactical systems. AIS algorithms, which control and present messages in order of priority in real time, are based on multiattribute characterization of tactical messages. The AIS employs an adaptive pattern recognition model to "learn" user preference structure incrementally during actual task performance. Across each of the command situations, the priority of messages is determined by the AIS in accord with the information selection behavior exhibited by the user in the model-training mode. The AIS was implemented and tested in a simulated environment. The results demonstrated the model's capability to 1) converge on distinctive information processing strategies exhibited by different operators; and 2) effectively present messages in their order of priority for each operator. The AIS is potentially useful in performing information distribution functions in command and control systems and in aiding the performance of personalized searches of large data bases.