CONTINGENCY HANDLING IN MISSION PLANNING FOR MULTI-ROBOT TEAMS by Shaurya Shriyam A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (Mechanical Engineering) December 2019 Copyright 2019 Shaurya Shriyam Dedication This dissertation is dedicated To my parents, Shreyash, school teachers, and college professors. ii Acknowledgements Doing a Ph.D. degree is an excellent academic journey, and I am very thankful to the University of Maryland, College Park, and the University of Southern California for providing me with an opportunity to experience it. The robotic labs I worked within both the universities had state-of-the-art robots and greatly stimulated my research progress. I would like to thank my advisor, Professor S. K. Gupta, for his constant motivation and terrific problem-solving skills, and guiding me through the challenges that I faced during my doctoral program. One faces the regular challenges during the completion of a doctoral program such as coming up with new ideas, producing results worthy of publication, and so on. My sincere appreciation goes to my supervisor for his excellent supervision during these five years. I would like to thank my committee members, Professor Azad Madni, and Professor Yan Jin, for serving in my committee. They provided helpful advice to improve my dissertation. Specifically, their guidance was useful for me to improve the structure and writing style of this dissertation I would like to acknowledge the financial support provided by the National Science Foundation (NSF) towards the completion of my research goals accumulating into this final dissertation. Although Robotic Smart Assistant for Manufacturing project is not part of this dissertation, it was jointly done with Professor Krishna Kaipa, and it was a great experience to work with the Baxter robot. I would like to thank all of my labmates with whom I interacted and worked during my research program. The discussions that I had with Brual, Ariyan, Pradeep, Shantanu, Rishi, Jason, Josh, iii Sarah, Aniruddha, Nithyanand, Akshay, and Srudeep significantly helped improve my research work. Finally, I would like to thank my family for their support across multiple continents, especially to my Mom who always had faith in me, and to my Dad who always gave me incredible words of encouragement, and to my Brother who took great interest in my research work and gave interesting feedback. iv Table of Contents Dedication ii Acknowledgements iii List Of Tables viii List Of Figures ix Abstract xii Chapter 1: Introduction 1 1.1 Motivation . .1 1.2 Goal and Scope . .5 1.3 Overview . .9 Chapter 2: Literature Review 11 2.1 Mission Planning Using Model Checking . 11 2.1.1 Multi-robot Mission Modeling for Model Checking . 11 2.1.2 Model Checking Tools for Verification and Evaluation . 16 2.1.3 Handling Stochastic Events . 21 2.1.4 Contingency Management . 23 2.2 Multi-robot Task Allocation . 24 2.3 Decomposition of Exploration Tasks . 32 Chapter 3: Modelling and Verification of Contingency Resolution Strategies for Multi-robot Missions Using Temporal Logic 42 3.1 Introduction . 42 3.2 Problem Statement . 44 3.3 Overview of Approach . 44 3.3.1 Nominal Mission Modeling for Deterministic Transitions . 44 3.3.2 Nominal Mission Modeling for Probabilistic Transitions . 47 3.3.3 Modeling of Contingency Resolution Strategies . 49 3.3.4 Model Checking and Verification . 51 3.3.4.1 Deterministic Transition Systems . 51 3.3.4.2 Probabilistic Transition Systems . 52 3.4 Task Network Generation . 54 3.5 Summary . 55 v Chapter 4: Multi-USV Case Study 56 4.1 Introduction . 56 4.2 Problem Statement . 57 4.3 Mission Modeling . 58 4.4 Model Checking and Verification . 71 4.5 Summary . 75 Chapter 5: Multi-robot Assembly Case Study 76 5.1 Introduction . 76 5.2 Problem Statement . 77 5.3 Overview of Approach . 77 5.4 Modeling of Assembly Cell . 81 5.4.1 Nominal Operation Description . 81 5.4.2 State Variables . 82 5.4.3 Transition Rules . 88 5.4.4 Nominal Operation Strategy . 89 5.4.5 Contingency Resolution Strategies . 90 5.5 Model Checking and Verification . 93 5.5.1 Run-time Analysis . 93 5.5.2 Assembly Cell Analysis . 95 5.5.3 Aggregate Analysis of Large-scale Assembly Operations . 99 5.6 Summary . 103 Chapter 6: Incorporation of Contingency Tasks in Nominal Task Allocation 104 6.1 Introduction . 104 6.2 Background . 110 6.2.1 Environment Modeling . 111 6.2.2 Resource Modeling . 112 6.2.3 Task Modeling . 113 6.2.3.1 Mission Tasks . 113 6.2.3.2 Contingency Tasks . 117 6.2.4 Nominal Mission Planning . 119 6.3 Problem Statement . 120 6.4 Task Schedule Optimization . 123 6.4.1 Task Execution by Partial Teams . 123 6.4.2 Handling Divisible Tasks . 126 6.4.3 Task Scheduling Strategies . 129 6.4.3.1 Heuristics Employed . 130 6.4.3.2 Task Allocation Without Task Division . 131 6.4.3.3 Task Allocation with Task Division . 133 6.4.4 Accounting for Uncertainty . 136 6.5 Incorporating Potential Contingency Tasks . 138 6.5.1 Background . 138 6.5.2 Search-space Pruning Heuristic . 141 6.5.3 Analysis of Pruning Heuristic . 142 6.5.4 Contingency Handling Algorithm . 143 6.6 Results and Illustrations . 146 6.6.1 Multi-robot Task Allocation . 146 6.6.2 Contingency Management . 156 6.7 Summary . 162 vi Chapter 7: Decomposition of Collaborative Surveillance Tasks with Uncertainty in Environmental Conditions and Robot Availability 164 7.1 Introduction . 164 7.2 Problem Formulation . 166 7.2.1 Definitions . 166 7.2.2 State Space Representation . 168 7.2.3 Problem Statement . 169 7.3 Overview of Approach . 169 7.3.1 Partitioning the Region of Interest . 170 7.3.2 Optimizing Partition . 179 7.4 Area Partitioning Using Known Velocity-map . 181 7.4.1 Approach . 181 7.4.2 Computational Results . 187 7.5 Area Partitioning Using Unknown Velocity-map . 190 7.5.1 Approach . 190 7.5.2 Computational Results . 191 7.6 Area Partitioning Under Variable Robot Availability . 193 7.7 Summary . 195 Chapter 8: Conclusions 196 8.1 Intellectual Contributions . 196 8.1.1 Modelling and Verification of Contingency Resolution Strategies for Multi- robot Missions . 197 8.1.2 Incorporation of Contingency Tasks in Nominal Task Allocation . 198 8.1.3 Decomposition of Collaborative Surveillance Tasks . 198 8.2 Anticipated Benefits . 198 8.3 Future Work . 199 Reference List 201 vii List Of Tables 4.1 Runtime statistics collected for different problem sizes . 74 5.1 Effect of initial part buffer on production time . 93 5.2 Effect of contingency resolution probability on production time . 93 5.3 Computational performance of s1 with varying parameters . 94 5.4 Computational performance of.