Dynamic electronic institutions in agent oriented cloud robotic systems
© Nagrath et al.; licensee Springer. 2015
Received: 31 July 2014
Accepted: 12 January 2015
Published: 1 March 2015
The dot-com bubble bursted in the year 2000 followed by a swift movement towards resource virtualization and cloud computing business model. Cloud computing emerged not as new form of computing or network technology but a mere remoulding of existing technologies to suit a new business model. Cloud robotics is understood as adaptation of cloud computing ideas for robotic applications. Current efforts in cloud robotics stress upon developing robots that utilize computing and service infrastructure of the cloud, without debating on the underlying business model. HTM5 is an OMG’s MDA based Meta-model for agent oriented development of cloud robotic systems. The trade-view of HTM5 promotes peer-to-peer trade amongst software agents. HTM5 agents represent various cloud entities and implement their business logic on cloud interactions. Trade in a peer-to-peer cloud robotic system is based on relationships and contracts amongst several agent subsets. Electronic Institutions are associations of heterogeneous intelligent agents which interact with each other following predefined norms. In Dynamic Electronic Institutions, the process of formation, reformation and dissolution of institutions is automated leading to run time adaptations in groups of agents. DEIs in agent oriented cloud robotic ecosystems bring order and group intellect. This article presents DEI implementations through HTM5 methodology.
KeywordsDynamic electronic institutions Cloud robotics Model driven engineering Cloud computing Peer-to-peer system Business model
A note to practitioners
Cloud computing is a business model for the internet. A typical scenario of cloud computing has a serving party that offers its infrastructure, platform or software resources to one or many clients across the network cloud. Cloud service businesses charge their clients based on the quality and volume parameters chosen as and when required by the client. Service contracts, banking and administrative mechanism created the trust envelop that made cloud computing business model a success. When we move the ideas of cloud computing to robotics, there are two kinds of adaptations that will take place. The first kind of adaptation will involve direct modification of cloud services to suit robotic applications while the second kind of adaptation will be on the lines of social and business ideas represented by cloud computing. We believe that this second kind of adaptation will require special tools and development methodologies. Cloud robotic entities include all robotic and non-robotic entities that collectively build a cloud robotic service ecosystem. Using software agents to represent cloud robotic entities will require minimal changes in the way those entities are independently developed by various vendors. Agents are also ideal for implementing social and business concerns of a cloud robotic entity. HTM5 (5 View Hyperactive Transaction Meta Model) is a 5-view meta-model for model driven development of agent oriented cloud robotic systems. The trade view of HTM5 promotes peer-to-peer exchange of services based on relationships and contracts between participating agents. Agents are autonomous entities with personal goals that may make them greedy in their interactions with other agents. Dynamic Electronic Institutions are modelled on the ideas of Institutions in Human societies. Norms based on trade contracts, social relationships and institutions bring a sense of order in multi agent systems. The aim of this article is to test feasibility of HTM5 methodology in implementing Dynamic Electronic Institutions.
Cloud computing is a relatively new business model for the Internet. NIST (National Institute of Standards and Technology- United States) defines cloud computing as "a pay-per-use model for enabling available, convenient, on-demand network access to a shared pool of configurable computing resources that can be rapidly provisioned and released with minimal management effort or service provider interaction." Cloud computing does not introduces a new computing or network technology but as a business model it remoulds the way existing technologies are used. Decreasing cost of internet connectivity and cheaper internet enabled devices has further improved the feasibility of cloud computing as a business model. Robotic researchers and engineers soon realized the benefits of cloud computing in robotics. Cloud based storage and processing expanded functionalities while carrying a minimal set of hardware on-board. Emergence of cloud robotics from cloud computing can be seen as a twofold development. The more visible development is direct modification of current cloud based services for robotic applications. Cloud robotics is a comprehensive term used to describe infrastructure, platform or software as a service for robots, internet enabled robotics, utilisation of search engines by robots and use of internet for communication between robots. These developments are making an impact in the way robotic systems are designed using cloud based tools but not much is done towards developing cloud robotics as a business model.
An agent oriented approach towards cloud robotics has some distinct advantages. A typical cloud robotic entity will have a manufacturer with a personalized development methodology for its product. The manufacturer would typically like to hide the internal designs of its product and thus the business that deploys that entity may have limited or no access to the internal software framework of the entity. The business will also want to keep its business logic hidden from the outside world and even from sections of their own workforce. “Software Agents are computational entities with specific roles and personal objectives working in a visible environment with other entities which may have dissimilar roles and objectives” (Jennings and Bussmann 2003). Using a software agent to represent cloud entities does not interfere with manufacturer’s development methodology. Agents are closed autonomous systems that have an internal logic framework that communicates with the outside world via messages. Unlike objects in an object oriented methodology, Agents do not release details of their functionalities and do not allow direct execution of their functions by other agents. Agents thus are by design ideal for implementing a secure business logic. Multi-agent systems (Luck et al. 2003, 2005; Wooldridge 2002) are also ideal for implementing intelligent concepts like Distributed Artificial Intelligence (DAI) (Stone and Veloso 2000) and digital business ecosystem (Discussed in Section 2, 2). An agent based approach is idea for systems with dynamic participation of entities in an open (Hungate and Gray 1995) peer-to-peer service exchange.
- 1.Names and Relative Locations of the following Trade elements in a cloud robotic ecosystem.
Components (Agent, Relation or Merge) that are involved in Trade.
Items in Trade.
Data entities associated with Trade items.
- 2.The following Information about the above entities:
Associations between Trade items and Components.
- (b)Nature of association between a Trade item and a Component:
Item is a Demand by a Component.
Item is a Service provided by a Component.
- (c)Entities associated with a Trade item:
Components that provide the item as a service.
Components which demand the item.
Data entities which are associated with Trade of the item.
The Components (Generally Relations) that are hosting and managing those data entities.
- (d)Nature of various Data entities:
Is it a Lookup table?
Is it a cost metric?
Is it a management variable?
- 3.Following Functionalities should exclusively go in the Relational view classes of various components:
Localization: Locating one’s position in different transactions.
Identifying relationships associated with a particular trade item.
Implementing relationship norms associated with a trade item.
Implementation of Business logic of a Component (Functionality related to business concerns of a particular HTM5 Component).
Implementation of Business logic of the system (Functionality related to business concerns of the cloud robotic system).
Calculating readjustments in relationship norms based on business logic.
Communicating desired readjustments to relational view classes.
Maintaining data entities associated with a trade item.
Reading and updating of remotely hosted data entities associated with a trade item.
Generating triggers for Trade Hyperactivity sub-view class. (Initiation, management or finalization of a Hyperactive link).
Dynamic electronic institutions
Human societies are amalgamation of several norm based institutions that give order to otherwise random interactions. Institutions are structures based on mutual incentives based on predefined contracts. Social, Political and economic institutions represents the norms of a society and interactions of its members. Institutes establish standardization in response from a member entity which in absence of an institution is free to act solely for its own benefit. Institution helps in controlling the greediness of individual entities and brings order to a system (North 1996). Electronic Institutions are a relatively new field where the concept of human institutions is extended to Multi Agent Systems (MAS) (Luck et al.2003,2005; Wooldridge2002) and Distributed Artificial Intelligence (DAI) (Stone and Veloso 2000). Some early attempts towards the use of organizational metaphors for system modelling systems were presented in (Pattison et al.1987; Werner1989). The first approach towards electronic institutions was given in (Noriega 1999) where an abstract notion agent-mediated electronic institution was introduced for the first time. These institutions are described as environments where agents are interacting with other agents under predefined restrictions. An institution is specified by a set of pre-defined norms that restrict actions of its member agents. The idea of an electronic institution is very open and various groups (Aldewereld et al.2005; Dignum2004; López 2003) are working on this problem with different perspectives. Electronic institutions require limited human intervention for institution design phase. In open agent systems, it is necessary to automate the design phase of institutions.
Formation: Agents with similar objectives come together to form a coalition. A coalition is usually not governed by a set of norms, but trust between agents may play a part in the coalition formation phase. Any logic that governs coalition formation between a set of agents is their Formation logic.
Re-Formation: Re-formation is the process of reconfiguring a coalition. A reformation may be triggered by change in coalition membership or a change in parameters responsible for coalition formation.
Foundation: The member agents in a coalition choose a candidate Institution to form. The norms of the candidate institution are based on collective views of individual members of the coalition. Once the target institution is selected, the coalition goes through institutionalization to form an institute of the selected type. This step presents a lot of challenges as selection of the kind of institution and maintenance of institution-base requires a well-defined strategy. In theory, any strategy that assigns an institute to a group of agents (based on their collective decision parameters) qualifies as a Foundation Logic.
Re-Foundation: Re-foundation is the process of reconfiguring an institution. A reconfiguration may be triggered by a change in environment variables or a foundation-timeout value set by Foundation logic (at the time of institution foundation).
Fulfilment: The member agents in an institute dissolve into individual free agents when the institute completes all its objectives. An institute may also fulfil when triggered by a fulfilment-timeout value set by Foundation logic (at the time of institution foundation). Like foundation, fulfilment is also a challenging logic to device. The decision may be based on weighted percentage of collective goals of member agents, or time elapsed since institution foundation / re-foundation. Fulfilment logic is usually devised at foundation state when the institute candidate is selected by Foundation logic.
Dynamic electronic institutions in cloud robotics
A Digital B2B electronic commerce ecosystem: Agent Community
A Business Opportunity: Coalition
Digital Business Ecosystem (DBE): Dynamic Institution
Search of opportunities (Formation Logic)
Analysis of opportunity by business owner (Validation I, Optional)
Coalition establishment (Formation and Re-Formation Phase, Re-formation will require re-validation)
DBE selection (Foundation Logic)
Acceptation by business owner (Validation II, Optional)
DBE establishment (Foundation and Re-Foundation Phase, Re-foundation will require re-validation)
DBE Finalization (Fulfilment Phase)
Dynamic electronic institutions in HTM5
In the previous section we saw the steps in a DBE lifecycle and its applicability in the cloud robotic domain. HTM5 (Nagrath et al.2013a,2013b) is OMG-MDA (OMG 2003) based meta-model for development of agent oriented peer-to-peer cloud robotic systems (See Section 2). The meta-model is designed to provide tools to implement advance distributed Artificial Intelligence (DAI) designs in a cloud robotic ecosystem. In this section we discuss the tools and anatomical elements of HTM5 that are utilized to implement a Digital Business Ecosystem based on Dynamic Electronic Institution.
For the example shown in Figure 6, the relations are designed to act as Institution seeds. No anatomical change is required in HTM5 to use HTM5 relation construct as Institution seeds. An institution between groups of agents can be seen as a special kind of relationship. In HTM5, the norms and relationship variables of a relationship are hosted and managed by the Relation agent. When HTM5 is used to implement Dynamic Electronic Institutions, the variables, formation and foundation logic may be hosted in institution seeds (which are relation agents). For ease of implementation, an institution may be implemented as two separate agents. In Figure 6, the Manager agent along with InsA relation and M4 merge are all hosted at one machine. It is possible to implement the logic of all three components (Manager, InsA and M4) onto one agent but separation and placement of machine functionalities into agent, merge and relation specific parts is encouraged in HTM5. In Figure 6 Part I of the figure is a full ARC diagram of the Sandbox system while part II is its normalized version. There is multiple numbers of Miner and Trader agents in the system. Part III models trade dependencies in Sandbox cloud ecosystem. Peer-to-Peer trade relationships exists between members of a relationship (Here relationships are modelled as Institutions). The Trade-Agent Relation Chart (Trade-ARC) that defines the following trade relationships and dependencies between members of Sandbox cloud robotic ecosystem.
Trade Search Space is a service provided by Trader to Manager and assigns a search space in the mine for the Manager.
Trade Sub Search Space is a service provided by Manager to Miners and assigns a section of the search space (allocated to the Manager) to the Miner.
Trade Sub Space Hit is the notification service from the Miner agent when it detects the target mineral. This service is a demand at Manager agent.
Trade Initial coordinates locates the found mineral in the mine field. This service is a demand at Trader agent.
Trade Miner Salary is the payment that a Miner agent receives in exchange of the mineral locations it delivers to the Trader agent. The service is a demand at the Bank agent which transfers the amount from Trader’s to Miner’s account.
Trade Cargo is a service by the Trader agent to the Market where it sells the acquired mineral locations.
Trade Cargo Payment is the payment that a Trader agent receives in exchange of the mineral locations (Cargo) it delivers to the Market agent. The service is a demand at the Bank agent which transfers the amount from Market’s to Trader’s account.
Demand Lookup Table Miner ID | Salary is a lookup table to get the desired salaries (Trade: Miner Salary) by individual Miner agents.
Demand Lookup Table Mineral | Price is a lookup table to get the prices of different minerals (Trade: Sub Space Hit) by individual Trader agents. A Miner agent chooses its target mineral based on the current price of minerals.
Demand Cost Metrice Mined Area Pc. Is a trade variable to check the percentage of mine’s area that is already explored. This is a metrice to know when to allocate a new search space (Trade: Sub-Search Space) to individual Miners.
Service Lookup Table Trader ID | Cargo is for the trade Cargo Payment and is used by the Market agent to initiate cargo payments.
Institution InsA is an institution between one Manager agent, one relation InsB, one Bank agent and Nm Miner agents. The institution manages the allocation, reallocation and management of mine spaces for individual Miners and allows for dynamic updating of a Miner’s asking salary based on inputs from Bank (Miner’s current bank balance) and Market Server (updated prices of different minerals).
Institution InsB is an institution between one Market agent, one relation InsA, one Bank agent and Nt Trader agents. The institution manages the allocation, reallocation and management of Cargo items for individual Traders and allows for dynamic updating of a Trader’s asking price for individual minerals based on inputs from Bank (Trader’s current bank balance) and President Machine (updated percentage of Mined area).
Run time Visualization of Market and Bank balances of agents in simulation exeriments: Figure 11 Part I shows sample graph of market values over a period of 5000 event steps. For every simulation run a unique random market pattern is generated that influences BOT behaviour. A randomiser seed is used to regenerate a particular market pattern. The results of running the experiment without institution formation are matched to the scenario where institutions can be formed. The profits of all BOTs for the market trend shown in Part I can be inspected at the end of the experiment as shown in Part II and III. The key observations from these experiments are explained in Section 2.
Figure 8 Part VII shows average profits of agents are always greater when they operate within institutions irrespective of the size of the agent community.
Figure 8 Part IX shows that the number of working groups is very high when agents do not operate with institutions. This is due to greedy nature on individual agents which force a working group to dissolve when their personal goals are not fulfilled. Institutions on the other hand are very low in number as they are dissolved when a collective decision is made.
Another observation from Figure 8 Part IX is that the number of working groups with institutions does not fluctuate a lot with changing market patterns (As institutes are less sensitive to minor fluctuations in market trend).
Figure 8 Part VIII shows that the working groups sustain for longer duration when they operate as institutions (More work hours per working group). The peak in Experiment 12 suggests that the man hours per working group go higher as experiments run for longer duration (Experiment 12 is the longest experiment).
With a few exceptions, an institution based ecosystem gives more profit per work hour (Figure 8 Part XI).
Conclusion and future direction
In this article we presented Dynamic Electronic Institutions (DEI) implementations in HTM5 meta-model for agent oriented development of cloud robotic systems. Digital Business Ecosystem (DBE) is one application domain of dynamic electronic institutions in cloud robotic colonies. HTM5 meta-model is designed for including distributed artificial intelligence designs on cloud robotic ecosystems. Peer to peer trade based on relationships between agents, representing heterogeneous cloud entities in the cloud using agents, an OMG-MDA based three layered design and its domain specificity makes HTM5 an ideal methodology for development of agent oriented cloud robotic systems. The case study, examples and discussions presented in the current article gives sufficient evidence that HTM5 is a feasible methodology for implementing complex trade and institution logics on a cloud robotic system. The complete HTM5 model, a domain specific language supporting HTM5 and a case study specific for peer to peer trade variability in HTM5 is currently submitted to well-known journals. The next step in this direction is to test the methodology in real life industrial projects and to improve the model by industrial feedback. A detailed user guide and a graphical design interface for HTM5 based designing is also currently under development.
The current research is being funded by the Laboratory Le2i (CNRS 6306, Le-Creusot, FRANCE), Bourgogne regional council (Regional French administration) and the Universiti Technologi Petronas (Perak, Malaysia).
- Aldewereld, H, Vazquez-Salceda J, Dignum F, Meyer JJC (2005) Norm compliance of protocols in electronic institutions In: Proceedings of the Fourth International Joint Conference on Autonomous Agents and Multiagent Systems, AAMAS ’05, ACM, New York, NY, USA, 1291–1292. doi:10.1145/1082473.1082738. http://doi.acm.org/10.1145/1082473.1082738.
- Clements, PC, Northrop L (2001) Software product lines: practices and patterns. Addison-Wesley Longman Publishing Co., Inc., Boston, MA, USA.Google Scholar
- Dignum, V (2004) A model for organizational interaction: based on agents, founded in logic. Ph.D. thesis, Universiteit Utrecht.Google Scholar
- Dynamic Electronic Institutions Digital Business Ecosystem and Peer to Peer Cloud Robotics Simulation Videos (2014). https://sites.google.com/site/deidbep2psimulations/.
- Finkelstein, A, Kramer J, Nuseibeh B, Finkelstein L, Goedicke M (1992) Viewpoints: a framework for integrating multiple perspectives in system development. Int J Software Eng Knowledge Eng 2(1): 31–57. http://dblp.uni-trier.de/db/journals/ijseke/ijseke2.html#FinkelsteinKNFG92.View ArticleGoogle Scholar
- Hungate, JI, Gray MM (1995) Conference report: application portability profile and open system environment users forum gaithersburg, md may 9-10, 1995. J Res Nat Inst Standards Technol 100(6): 699–709.View ArticleGoogle Scholar
- Jennings, NR, Bussmann S (2003) Agent-based control systems. Why are they suited to engineering complex systems?. IEEE Control Syst Mag 23(3): 61–73.View ArticleGoogle Scholar
- López, FLY (2003) Social power and norms: impact on agent behaviour. Tech. rep.Google Scholar
- Luck, M, McBurney P, Preist C (2003) Agent technology: enabling next generation computing (a roadmap for agent based computing). AgentLink.Google Scholar
- Luck, M, McBurne P, Shehory O, Willmott S (2005) Agent technology: computing as interaction (a roadmap for agent based computing). AgentLink.Google Scholar
- Muntaner-Perich, E, Rosa Esteva J (2007) Using dynamic electronic institutions to enable digital business ecosystems. In: Noriega P, Vzquez-Salceda J, Boella G, Boissier O, Dignum V, Fornara N, Matson E (eds)Coordination, Organizations, Institutions, and Norms in Agent Systems II, Lecture Notes in Computer Science, 259–273.. Berlin, Germany, Springer Berlin Heidelberg.View ArticleGoogle Scholar
- Muntaner-Perich, E, Rosa Esteva J (2008) Towards a formalisation of dynamic electronic institutions. In: Sichman J, Padget J, Ossowski S, Noriega P (eds)Coordination, Organizations, Institutions, and Norms in Agent Systems III, Lecture Notes in Computer Science, 97–109.. Berlin, Germany, Springer Berlin Heidelberg.View ArticleGoogle Scholar
- Nagrath, V, Meriaudeau F, Malik AS, Morel O (2012) Agent relation charts (ARCs) for modeling cloud based transactions In: Proceedings of the 2012 International Conference on Communication Systems and Network Technologies, CSNT ’12, 704–709.. IEEE Computer Society, Washington, DC, USA. doi:10.1109/CSNT.2012.156. http://dx.doi.org/10.1109/CSNT.2012.156.View ArticleGoogle Scholar
- Nagrath, V, Meriaudeau F, Malik A, Morel O (2013a) Introducing the concept of hyperactivity in multi agent systems In: Communication Systems and Network Technologies (CSNT), 2013 International Conference on, conference CSNT2013, 542–546.. MIR Labs Thatipur, Gwalior, India. doi:10.1109/CSNT.2013.117.Google Scholar
- Nagrath, V, Morel O, Malik A, Saad N, Meriaudeau F (2013b) HTM5-Trade model for relationship based trade modelling in multi agent systems In: Science and Information Conference (SAI), 2013, 188–196.Google Scholar
- Noriega, P (1999) Agent mediated auctions: the fishmarket metaphor. Institut d’Investigació, en Intel· ligència Artificial.Google Scholar
- North, DC (1996) Economics and cognitive Science. Economic History 9612002, EconWPA. http://ideas.repec.org/p/wpa/wuwpeh/9612002.html.
- OMG (2003) Model driven architecture (MDA) guide. OMG doc. ab/2003-06-01 http://www.omg.org/mda/.
- Pattison, HE, Corkill DD, Lesser VR (1987) Instantiating descriptions of organizational structures In: MN Huhns (ed.) Distributed Artificial Intelligence, Research Notes in Artificial Intelligence, chap. 3, 59–96, Pitman, 6 East 43rd Street, New York, NY, 1001, 7 USA.Google Scholar
- Pohl, K, Böckle G, Linden FJvd (2005) Software product line engineering: foundations, principles and techniques. Springer-Verlag New York, Inc., Secaucus, NJ, USA.View ArticleGoogle Scholar
- Seidewitz, E (2003) What models mean. IEEE Softw 20(5): 26–32. doi:10.1109/MS.2003.1231147 http://dx.doi.org/10.1109/MS.2003.1231147.View ArticleGoogle Scholar
- Stone, P, Veloso M (2000) Multiagent systems: a survey from a machine learning perspective. Autonomous Robots 8(3): 345–383. doi:10.1023/A:1008942012299 http://www.cs.cmu.edu/~mmv/papers/MASsurvey.pdf.
- TurtleBOTs. [Latest access on 06/11/2014 0344 Hrs] http://www.turtlebot.com/.
- Visual Basic for Applications (2014). [Latest access on 06/11/2014 0344 Hrs] http://excelvbatutor.com.
- VisualBots Simulator (2014). [Latest access on 06/11/2014 0344 Hrs] http://www.visualbots.com/.
- Weiss, DM, Lai CTR (1999) Software product-line engineering: a family-based software development process. Addison-Wesley Longman Publishing Co., Inc., Boston, MA, USA.Google Scholar
- Weiss, G (1999) Multiagent Systems: A Modern Approach to Distributed Artificial Intelligence. MIT Press, Cambridge, MA, USA.Google Scholar
- Werner, E (1989) Cooperating agents: a unified theory of communication and social structure. In: Gasser IL Huhns MN (eds)Distributed Artificial Intelligence (Vol. II), 3–36, Kaufmann, San Mateo, CA.Google Scholar
- Wooldridge, M (2002) An introduction to multi-agent systems Ames, IA 50010. (Corporate Headquarters of John Wiley & Sons), USA.Google Scholar
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited.