Route Planning and Operator Allocation in Robot Fleets
MetadataShow full item record
In this thesis, we address various challenges related to optimal planning and task allocation in a robot fleet supervised by remote human operators. The overarching goal is to enhance the performance and efficiency of the robot fleets by planning routes and scheduling operator assistance while accounting for limited human availability. The thesis consists of three main problems, each of which focuses on a specific aspect of the system. The first problem pertains to optimal planning for a robot in a collaborative human-robot team, where the human supervisor is intermittently available to assist the robot to complete its tasks faster. Specifically, we address the challenge of computing the fastest route between two configurations in an environment with time constraints on how long the robot can wait for assistance at intermediate configurations. We consider the application of robot navigation in a city environment, where different routes can have distinct speed limits and different time constraints on how long a robot is allowed to wait. Our proposed approach utilizes the concepts of budget and critical departure times, enabling optimal solution and enhanced scalability compared to existing methods. Extensive comparisons with baseline algorithms on a city road network demonstrate its effectiveness and ability to achieve high-quality solutions. Furthermore, we extend the problem to the multi-robot case, where the challenge lies in prioritizing robots when multiple service requests arrive simultaneously. To address this challenge, we present a greedy algorithm that efficiently prioritizes service requests in a batch and has a remarkably good performance compared to the optimal solution. The next problem focuses on allocating human operators to robots in a fleet, considering each robot's specified route and the potential for failures and getting stuck. Conventional techniques used to solve such problems face scalability issues due to exponential growth of state and action spaces with the number of robots and operators. To overcome these, we derive conditions for a technical requirement called indexability, thereby enabling the use of the Whittle index heuristic. Our key insight is to leverage the structure of the value function of individual robots, resulting in conditions that can be easily verified separately for each state of each robot. We apply these conditions to two types of transitions commonly seen in supervised robot fleets. Through numerical simulations, we demonstrate the efficacy of Whittle index policy as a near-optimal scalable approach that outperforms existing scalable methods. Finally, we investigate the impact of interruptions on human supervisors overseeing a fleet of robots. Human supervisors in such systems are primarily responsible for monitoring robots, but can also be assigned with secondary tasks. These tasks can act as interruptions and can be categorized as either intrinsic, i.e., being directly related to the monitoring task, or extrinsic, i.e., being unrelated. Through a user study involving $39$ participants, the findings reveal that task performance remains relatively unaffected by interruptions, and is primarily dependent on the number of robots being monitored. However, extrinsic interruptions led to a significant increase in perceived workload, creating challenges in switching between tasks. These results highlight the importance of managing user workload by limiting extrinsic interruptions in such supervision systems. Overall, this thesis contributes to the field of robot planning and operator allocation in collaborative human-robot teams. By incorporating human assistance, addressing scalability challenges, and understanding the impact of interruptions, we aim to enhance the performance and usability of robot fleets. Our work introduces optimal planning methods and efficient allocation strategies, empowering the seamless operation of robot fleets in real-world scenarios. Additionally, we provide valuable insights into user workload, shedding light on the interactions between humans and robots in such systems. We hope that our research promotes the widespread adoption of robot fleets and facilitates their integration into various domains, ultimately driving advancements in the field.
Cite this version of the work
Abhinav Dahiya (2023). Route Planning and Operator Allocation in Robot Fleets. UWSpace. http://hdl.handle.net/10012/20022
Showing items related by title, author, creator and subject.
Evaluation of Laban Effort Features based on the Social Attributes and Personality of Domestic Service Robots Emir, Ebru (University of Waterloo, 2022-09-29)Today, it is not uncommon to see robots adopted in various domains and environments. From manufacturing facilities to households, robots take over several roles and tasks. For instance, the adoption of robotic vacuum ...
Wu, Xianghai (University of Waterloo, 2009-10-28)Current methods of robot task teaching and learning have several limitations: highly-trained personnel are usually required to teach robots specific tasks; service-robot systems are limited in learning different types of ...
Baghaei Ravari, Parastoo (University of Waterloo, 2021-01-26)Social robots used in education can take different roles, including tutor robots and peer robots. Peer robots (also called teachable robots) take the role of a novice in a teaching interaction while the students take the ...