Dybalova, Daniela (2017) Flexible autonomy and context in human-agent collectives. PhD thesis, University of Nottingham. Access from the University of Nottingham repository: http://eprints.nottingham.ac.uk/43397/1/Thesis.pdf Copyright and reuse: The Nottingham ePrints service makes this work by researchers of the University of Nottingham available open access under the following conditions. This article is made available under the University of Nottingham End User licence and may be reused according to the conditions of the licence. For more details see: http://eprints.nottingham.ac.uk/end_user_agreement.pdf For more information, please contact [email protected] Flexible Autonomy and Context in Human-Agent Collectives DANIELA DYBALOVA THESIS SUBMITTED TO THE UNIVERSITY OF NOTTINGHAM FOR THE DEGREE OF DOCTOR OF PHILOSOPHY JUNE, 2017 -ii- ubi lex, ibi poena -iii- -iv- Abstract Human-agent collectives (HACs) are collaborative relationships between humans and software agents that are formed to meet the individual and collective goals of their members. In general, different members of a HAC should have differing degrees of autonomy in determining how a goal is to be achieved, and the degree of autonomy that should be enjoyed by each member of the collective varies with context. This thesis explores how norms can be used to achieve context sensitive flexible autonomy in HACs. Norms can be viewed as defining standards of ideal behaviour. In the form of rules and codes, they are widely used to coordinate and regulate activity in human organisations, and more recently they have also been proposed as a coordination mechanism for multi-agent systems (MAS). Norms therefore have the potential to form a common frame- work for coordination and control in HACs. The thesis develops a novel framework in which group and individual norms are used to specify both the goal to be achieved by a HAC and the degree of autonomy of the HAC and/or of its members in achieving a goal. The framework allows members of a collective to create norms specifying how a goal should (or should not) be achieved, together with sanctions for non-compliance. These norms form part of the decision making context of both the hu- mans and agents in the collective. A prototype implementation of the framework was evaluated using the Colored Trails test-bed in a scenario involving mixed human-agent teams. The experiments confirmed that norms can be used for coordination of HACs and to facilitate context related flexible autonomy. -v- -vi- Acknowledgements My first and greatest debt of gratitude must go to my advisors, Brian Lo- gan and Tom Rodden. Who patiently guided me through the challenges and difficulties that I faced. In particular Brian’s way of thinking he kept my ideas on track, and his enthusiasm and feedback were a big part of making it through. I would like thank Wenchao Jiang for making the code of the Geo- Sense game available and for assistance in developing the gameserver middleware. Special credit goes to Bas Testerink for this help with the integration of 2OPL. This work was funded by EPSRC grant EP/I011587/1. -vii- -viii- Contents 1 Introduction1 1.1 Problem Definition and Objectives..............2 1.2 Approach.............................3 1.3 Contribution...........................4 1.4 Structure of the Thesis.....................5 2 Literature Review7 2.1 Agents..............................7 2.1.1 Autonomy........................8 2.1.2 Flexible Autonomy................... 11 2.2 Human-Agent Interaction................... 15 2.3 Norms.............................. 17 2.3.1 Normative Multi-Agent Systems........... 18 2.3.2 Normative Programming Frameworks....... 20 2.3.2.1 Applying Norms to Games......... 22 2.3.2.2 Norms in Operational Multi-Agent Systems..... 23 2.3.2.3 Norm Monitoring and Enforcement.... 24 2.4 Programming Norm-Aware Agents............. 25 2.4.1 Norm-Aware Deliberation............... 31 2.4.2 Norm Representation................. 33 2.4.3 From Norms to Flexible Autonomy......... 35 2.5 Towards Group Norms..................... 36 2.5.1 Team work: Joint intentions and SharedPlans theory....... 37 2.5.2 Agency theory..................... 40 2.5.3 Collective responsibility................ 40 2.5.4 Sanctioning in a group................. 42 2.6 Summary............................. 43 3 NormHACing Framework 45 3.1 Introduction........................... 47 3.1.1 Contributions...................... 48 3.2 2APL............................... 48 -ix- CONTENTS 3.2.1 Beliefs and Goals.................... 49 3.2.2 Basic actions....................... 50 3.2.3 Plans........................... 52 3.2.4 Practical reasoning rules................ 52 3.2.5 External environments................. 55 3.2.6 Events and exceptions................. 56 3.2.7 2APL files........................ 56 3.3 N-2APL.............................. 57 3.3.1 Beliefs, Goals and Events............... 57 3.3.2 Actions & Plans..................... 58 3.3.3 Implementation..................... 61 3.4 2OPL............................... 64 3.5 Middle-ware........................... 68 3.6 Application........................... 70 3.6.1 Game Play........................ 71 3.6.2 Encoding Game Rules as Norms........... 73 3.6.3 Agent Programs.................... 74 3.6.4 Gameserver Integration................ 75 3.6.5 Evaluation........................ 76 3.7 Summary............................. 76 4 Group Norms and Human-Agent Collectives 79 4.1 Introducing Group Norms................... 80 4.2 Taxonomy of Group Norms.................. 81 4.3 Extensions of Aldewereld et al................ 85 4.4 Formalising Group Norms................... 88 4.5 Team Plan............................ 89 4.5.1 Sanctioning Policy................... 90 4.6 Responsible Agent....................... 91 4.7 Hierarchical Group Norms................... 92 4.8 Implications for Agents..................... 93 4.8.1 Human Agents..................... 93 4.8.2 Software Agents.................... 95 4.8.3 Flexible Autonomy with Group Norm........ 96 4.9 Examples............................. 96 4.9.1 Birthday Example................... 96 4.9.2 Location based game example............ 97 4.10 Summary............................. 98 5 NormHACing+ 101 5.1 Motivation............................ 102 5.2 G-2OPL.............................. 102 5.2.1 Syntax.......................... 103 5.2.2 Groups.......................... 104 5.2.3 Sanctions......................... 106 5.2.4 Execution........................ 106 5.2.4.1 Fact Base Updates.............. 107 5.2.4.2 Norm Processing............... 107 -x- CONTENTS 5.2.4.3 Application of Sanctions.......... 108 5.2.5 Simplification of the Hierarchical Norms...... 109 5.3 GN-2APL............................. 109 5.3.1 Limitations....................... 110 5.4 Middle-ware........................... 111 5.5 Human Agents......................... 112 5.6 Applying the Framework to a Location Based Game................... 112 5.7 Summary............................. 113 6 Evaluation 115 6.1 Evaluation Questions...................... 116 6.2 Methodology of The Evaluation................ 118 6.2.1 Colored Trails...................... 118 6.2.1.1 Game Play.................. 120 6.2.1.2 Interface.................... 120 6.2.2 Norms for Human Players.............. 121 6.2.3 Game Set-Up...................... 122 6.2.3.1 Standard Game............... 123 6.2.3.2 Flexible Game................ 126 6.3 Study I. Norms as Coordination Mechanism........ 128 6.3.1 Results.......................... 129 6.3.1.1 Question 1: Norms as Coordination Mech- anism..................... 129 6.3.1.2 Question 2: Flexible Autonomy...... 136 6.3.1.3 Comments Summary............ 137 6.4 Study II. Focus on Extended Flexible Autonomy................. 141 6.4.1 Results.......................... 142 6.4.1.1 Question 1: Norms as Coordination Mech- anism..................... 143 6.4.1.2 Question 2: Flexible Autonomy...... 147 6.4.1.3 Comments Summary............ 148 6.5 Additional Findings...................... 149 6.6 Reflections on Norms in HACs................ 150 6.7 Summary............................. 151 7 Conclusions 153 7.1 Contributions.......................... 155 7.2 Future Work........................... 156 7.2.1 Human-Agent Interaction............... 157 7.2.2 Norm-Aware Agents.................. 158 7.3 Publications of the Author................... 161 Appendices A Appendix 165 B Appendix 169 -xi- CONTENTS C Appendix 177 D Appendix 183 E Appendix 185 References 210 -xii- List of Figures 2.1 Example of obligation..................... 34 2.2 Example of prohibition..................... 35 2.3 Literature review topics.................... 44 3.1 Overall system architecture.................. 49 3.2 Example of agent’s beliefs................... 50 3.3 Example of agent’s goals.................... 50 3.4 EBNF syntax of N-2APL.................... 61 3.5 N-2APL: Scheduling Algorithm................ 65 3.6 EBNF of a 2OPL norm..................... 66 3.7 Example GeoSense game norms................ 66 3.8 Example GeoSense game norms................ 67 3.9 JavaSpace application example................ 69 3.10 Entry objects for obligations and prohibitions........ 70 3.11 GeoSense web interface.................... 71 3.12 An example of a HTTP request and a response in JSON format.............................. 72 3.13 Example GeoSense game norms................ 73 3.14 N-2APL program for the Truck agent............. 74 3.15 Overall system architecture.................. 75 4.1 Example of group obligation.................