Swarup Mohalik - Academia.edu (original) (raw)

Papers by Swarup Mohalik

arXiv (Cornell University), Dec 18, 2018

Intelligent Cyber-physical systems can be modeled as multi-agent systems with planning capability... more Intelligent Cyber-physical systems can be modeled as multi-agent systems with planning capability to impart adaptivity for changing contexts. In such multi-agent systems, the protocol for plan execution must result in the proper completion and ordering of actions in spite of their distributed execution. However, in untrusted scenarios, there is a possibility of agents not respecting the protocol either due to faults or due to malicious reasons thereby resulting in plan failure. In order to prevent such situations, we propose to implement the execution of agents through smart contracts. This points to a generic architecture seamlessly integrating intelligent planning-based CPS and smartcontracts.

Realization of industry-scale, goal-driven, autonomous systems with AI planning technology faces ... more Realization of industry-scale, goal-driven, autonomous systems with AI planning technology faces several challenges: flexibly specifying planning goal states in varying situations, synthesizing plans in large state spaces, re-planning in dynamic situations, and facilitating humans to supervise, give feedback and intervene. In this paper, we present Intent-driven Strategic Tactical Planning (ISTP) to address these challenges. We demonstrate its efficacy through its application for radio base station inspection across several locations using drones. The inspection task involves capturing images, thermal images or signal measurements-called knowledge-objects-of various components of the base stations for downstream processing. In the ISTP approach, an operator indicates her goals by flying the drone to different components of interest. These goals are generalized to capture the intent of the operator, which are then instantiated in new situations to generate goals dynamically. Towards planning and re-planning in large state spaces to achieve these goals efficiently, we extend the Strategic-Tactical Planning paradigm. All the components of ISTP are integrated in an intuitive UI and demonstrated through a real life use-case built on the UNITY simulator platform.

As IoT devices proliferate, platforms and programming environments to develop IoT-based systems a... more As IoT devices proliferate, platforms and programming environments to develop IoT-based systems are becoming commonplace. However, the current models of development will soon prove to be inadequate due to the exploding scale, variety and dynamism in the IoT ecosystems, which is making it imperative that these systems manage and operate themselves in an autonomous fashion. Specifically, IoT-based systems must be able to adapt themselves intelligently to changes in the device hardware and software, the context and context-dependent policies and continue delivering to the requirements. Unfortunately, current IoT platforms and programming environments do not have any native support for such intelligence. In order to address this lacuna, we suggest additional components and APIs that can support intelligent autonomy based on the MAPE-K (Monitor, Analyze, Plan, Execute, Knowledge) architecture. The solution is demonstrated through a couple of concrete case studies implemented using IoT sensors and actuators on Raspberry Pi boards, openHAB - a popular IoT automation environment, Metric-FF - a well-known search-based AI planner and Leshan, an LwM2M platform for providing the sensing and actuation interfaces of the IoT devices.

ACM Sigsoft Software Engineering Notes, Jan 11, 2018

With the proliferation of the Internet of Things and the associ- ated trend of integration of sof... more With the proliferation of the Internet of Things and the associ- ated trend of integration of software with "things", it is predicted that the complexity of the systems-of-future will be characterized not only by scale and variety of devices and software but also by the constant change of the system context due to the mobil- ity of devices and M2M interactions. Consequently, the current paradigm of automation will be inadequate for the management and operation of these systems. The dominant approach to ad- dress this issue is to design and develop autonomous systems that can adapt to the changes in various levels and keep delivering the expected functionality. Such systems need fundamentally different architectures, components and methodologies incorporating new paradigms such as machine learning and intelligent decision making. In this workshop, we attempt to discuss the in uence of these paradigms on the development life cycle of adaptive soft- ware, starting from requirements, design and architecture and also their verification and validation.

ArXiv, 2020

As a field of AI, Machine Reasoning (MR) uses largely symbolic means to formalize and emulate abs... more As a field of AI, Machine Reasoning (MR) uses largely symbolic means to formalize and emulate abstract reasoning. Studies in early MR have notably started inquiries into Explainable AI (XAI) -- arguably one of the biggest concerns today for the AI community. Work on explainable MR as well as on MR approaches to explainability in other areas of AI has continued ever since. It is especially potent in modern MR branches, such as argumentation, constraint and logic programming, planning. We hereby aim to provide a selective overview of MR explainability techniques and studies in hopes that insights from this long track of research will complement well the current XAI landscape. This document reports our work in-progress on MR explainability.

2017 IEEE Symposium on Service-Oriented System Engineering (SOSE), 2017

Service-oriented architecture (SOA) for Cyber PhysicalSystems (CPS) can be implemented through th... more Service-oriented architecture (SOA) for Cyber PhysicalSystems (CPS) can be implemented through the abstractionof sensing and actuation capabilities of devices as microservicesand providing data processing and decision-making services athigher levels. However, specific to CPS, the critical requirementof adaptivity must be taken into account so that thearchitectures can autonomously handle the dynamic changes inthe devices and their environment making them transparent tothe applications. Towards this, we implement the well-knownMAPE-K (Monitor, Analyze, Plan, Act, using stored Knowledge)reference architecture from autonomic computing, with the corePlan function powered by AI planning techniques. The proof ofconcept prototype is built upon our earlier works: InteropAdapt, a middleware for interoperability of control actions between theapplication layer and the device layer, and HINC, an informationmodel that harmonizes IoT resources spanning devices, networkfunctions and cloud resources. We illustrate our ideas via realisticexamples in the building automation domain.

arXiv (Cornell University), Feb 26, 2018

Antifragile systems grow measurably better in the presence of hazards. This is in contrast to fra... more Antifragile systems grow measurably better in the presence of hazards. This is in contrast to fragile systems which break down in the presence of hazards, robust systems that tolerate hazards up to a certain degree, and resilient systems that-like selfhealing systems-revert to their earlier expected behavior after a period of convalescence. The notion of antifragility was introduced by Taleb for economics systems, but its applicability has been illustrated in biological and engineering domains as well. In this paper, we propose an architecture that imparts antifragility to intelligent autonomous systems, specifically those that are goal-driven and based on AI-planning. We argue that this architecture allows the system to self-improve by uncovering new capabilities obtained either through the hazards themselves (opportunistic) or through deliberation (strategic). An AI planning-based case study of an autonomous wheeled robot is presented. We show that with the proposed architecture, the robot develops antifragile behaviour with respect to an oil spill hazard.

In a Software Product Line (SPL) comprising specifications (feature sets), implementations (compo... more In a Software Product Line (SPL) comprising specifications (feature sets), implementations (component sets) and traceability between them, the definition of product is quite subtle. Intuitively, a strong relation of implementability should be established between implementations and specifications due to traceability. Various notions of traceability has been proposed in the literature : [13], [17], [8], [9]; but we found in our experience that they do not capture all situations that arise in practice. One example is the case where, an implementation, due to packaging reasons, contains additional components not required for a particular product specification. We have defined a general notion of traceability in order to cover such situations. Moreover, state-of-the-art satisfiability based notions lead to products where the implementability relation does not exist. Therefore, in this paper, we propose a simple, set-theoretic formalism to express the notions of traceability and implementability in a formal manner. The subsequent definition of SPL products is used to introduce a set of analysis problems that are either refinements of known problems, or are completely novel. Last but not the least, we propose encoding the analysis problems as Quantified Boolean Formula (QBF) constraints and use Quantified SAT (QSAT) solvers to solve these problems efficiently. To the best of our knowledge, the QBF encoding is novel; we prove the correctness of our encoding and demonstrate its practical feasibility through our prototype implementation Software Product Line Engine (SPLE).

5G networks introduce unprecedented flexibility and dynamic adaptation into service delivery and ... more 5G networks introduce unprecedented flexibility and dynamic adaptation into service delivery and network resource utilization. In the business layer, this is reflected in the ability to offer customizable service products with detailed agreements on functional and non-functional characteristics as well as fast delivery. Dynamic adaptation to changes within the constraints of stringent requirements on lead and reaction times is beyond the capacity of a human workforce. Extensive automation will be necessary to overcome this challenge.

Tele-operated and tele-presence robotics have been gaining industrial traction due to the advanta... more Tele-operated and tele-presence robotics have been gaining industrial traction due to the advantages of remote operation and flexibility. Generating network slices for Industry 4.0 tele-operation has traditionally been a static process with redundant network resources and spectrum over provisioning. However, this adds constraints on scalability, resource provisioning and costs in dedicated industrial networks. In this work, we explore the use of Artificial Intelligence Planning techniques to jointly plan, schedule and reconfigure robotic tasks in conjunction with appropriate network slicing. Through detailed analysis of resources and process timelines involved in slice provisioning and scaling, we demonstrate provisioning of appropriate slices with respect to robotic tasks. This is evaluated over a case study involving tele-operated robots for remote inspection in Industry 4.0 manufacturing. The system is able to reconfigure tasks in an appropriate manner to reduce end-to-end processing delays, while using just a fraction of network resources.

SAE Technical Paper Series, Apr 16, 2012

Attention: This paper is not yet published. If you are interested in purchasing it-via Mail, Fax ... more Attention: This paper is not yet published. If you are interested in purchasing it-via Mail, Fax or Download-please click on the" Mail" shopping cart icon below and proceed through the checkout process once you are finished browsing our web site. When this paper becomes ...

SAE International Journal of Passenger Cars - Electronic and Electrical Systems, Apr 8, 2013

Well-connected, regulatory compliant flight paths are crucial for UAVs to be adopted in mission-c... more Well-connected, regulatory compliant flight paths are crucial for UAVs to be adopted in mission-critical applications. In this paper, we present the Connectivity-Aware Path plannEr with Regulatory compliance (CAPER): a solution for planning safe, cellular-connected UAV paths in environments with heterogeneous connectivity regions, such that the planned paths comply with regulatory no-fly zones and height constraints. CAPER builds on the sampling-based planner Rapidly-exploring Random Trees (RRT), and makes a number of algorithmic modifications both in the planner and the collision detector. RRT has seen widespread use in planning paths in robotics, due to its ability to quickly search high dimensional spaces for feasible paths. However, several challenges exist in adopting RRTs for the connectivity-aware path planning problem in realistic spaces, which CAPER seeks to alleviate. In this paper we detail CAPER, and present results of its implementation in two realistic urban environments in Stockholm and Los Angeles. Since CAPER is built on the randomized algorithm RRT, we also present a brief analysis of multiple runs within the same environment.

Software Product Line (SPL) is a software development framework to jointly design a family of clo... more Software Product Line (SPL) is a software development framework to jointly design a family of closely related software products in an efficient and cost-effective manner. In order to separate the concerns and handle complexity, designers usually project the SPL along different perspectives such as feature, architecture and behaviour. Each perspective deals with variability of a set of artifacts and variability constraints among them. SPL designers attempt to ensure the consistency of the individual perspectives and the SPL as a whole. They are also interested in finding the elements common to all products and the live elements (used in at least one product). In the literature, most of the works focus on a single perspective and address the above-mentioned problems within single perspectives. There have also been attempts to express the variability of different perspectives within the feature perspective. However, since the different perspectives have different intents, coercing them into a single perspective may result in unnatural constructs in the feature perspective. Hence, it is better to keep the perspectives separate. However, in any SPL, the perspectives are closely related through an implementability relation or through constraints arising from design or business reasons. We call this the traceability aspect, which mandates an integrated analysis of the different perspectives. In this paper, we propose a constraint-based framework where variability and traceability constraints can be uniformly expressed, at the same time keeping the different intents of perspectives intact. We describe how the consistency, liveness, and commonness problems can be reduced to problems of constraint solving. Through a realistic case study, we provide some evidence that the constraintbased framework is expressive and scalable to large SPLs.

The Internet of Things (IoT) phenomenon is giving rise to large scale IoT deployments comprising ... more The Internet of Things (IoT) phenomenon is giving rise to large scale IoT deployments comprising thousands of IoT devices with large degree of heterogeneity. The heterogeneity is further aggravated by dynamism-new applications and requirements, changes in capabilities of devices during their life-cycle and mobility of devices and applications. This is leading to semantic interoperability issues among these devices, which is hindering the ability of system designers to draw the maximum value from these deployments. To address this problem, we present InteropAdapt, an adaptive middleware that can maintain seamless semantic interoperability across dynamic events. The central aim of this paper is to exhibit different architectural elements of InteropAdapt in some detail. We also illustrate InteropAdapt via a simple yet realistic example in the smart office automation domain.

In this paper, we suggest a service-oriented architecture for planning and execution (SOA-PE) in ... more In this paper, we suggest a service-oriented architecture for planning and execution (SOA-PE) in large scale cyberphysical systems (CPS). SOA-PE provides a clean separation between domain modeling, planning, execution, monitoring and actuation services. This approach helps realize the system-ofsystems paradigm allowing the decomposition of system goals into smaller subgoals, thus enhancing the scalability of the proposed solution. In addition to supporting large scale, autonomous systems, the service-oriented approach provides several benefits such as reusability, independent development and deployment, platform independence, transparency and flexibility, to the core services of Planning and Execution in these systems. The architecture targets decentralized, multi-agent systems for solutions like smart transportation and logistics and can scale to larger IoT use cases like smart cities. We illustrate the functionalities of the architecture through a prototype implementation and a case study from the logistics domain.

ABSTRACT Completely automatic generation of tests from formal executable test models of industria... more ABSTRACT Completely automatic generation of tests from formal executable test models of industrial size still looks like a “holy grail”, in spite of significant progress in model-based testing research and tool development. Realizing this, we follow a more down-to-earth approach by assuming that, even if a test model is available, the test expert manually derives powerful test fragments and what remains to be automated is chaining them into an optimal test. Focusing on this task, we develop a test optimization framework using an FSM extended with input variables and clocks, which reflects important features of Simulink/Stateflow statecharts. The test optimization is expressed as the Asymmetric Travelling Salesman Problem (ATSP). We show how this approach can be used for solving some testing problems specific to automotive controllers. We describe a proof-of-concept prototype, implementing the proposed approach, which we tested on a case study of a particular controller available along with some tests. Experiments with the prototype indicate that the approach scales well for hundreds of tests.

In this paper, we address the web service selection problem for linear workflows. Given a linear ... more In this paper, we address the web service selection problem for linear workflows. Given a linear workflow specifying a set of ordered tasks and a set of candidate services providing different features for each task, the selection problem deals with the objective of selecting the most eligible service for each task, given the ordering specified. A number of approaches to solving the selection problem have been proposed in literature. With web services growing at an incredible pace, service selection at the Internet scale has resurfaced as a problem of recent research interest. In this work, we present our approach to the selection problem using an abstraction refinement technique to address the scalability limitations of contemporary approaches. Experiments on web service benchmarks show that our approach can add substantial performance benefits in terms of space when compared to an approach without our optimization.

Journal of Logic, Language and Information, Jan 13, 2010

We suggest that developing automata theoretic foundations is relevant for knowledge theory, so th... more We suggest that developing automata theoretic foundations is relevant for knowledge theory, so that we study not only what is known by agents, but also the mechanisms by which such knowledge is arrived at. We define a class of epistemic automata, in which agents' local states are annotated with abstract knowledge assertions about others. These are finite state agents who communicate synchronously with each other and information exchange is 'perfect'. We show that the class of recognizable languages has good closure properties, leading to a Kleene-type theorem using what we call regular knowledge expressions. These automata model distributed causal knowledge in the following way: each agent in the system has a partial knowledge of the temporal evolution of the system, and every time agents synchronize, they update each other's knowledge, resulting in a more up-to-date view of the system state. Hence we show that these automata can be used to solve the satisfiability problem for a natural epistemic temporal logic for local properties. Finally, we characterize the class of languages recognized by epistemic automata as the regular consistent languages studied in concurrency theory.