Generalized task allocation and route planning for robots with multiple depots in indoor building environments

Highlights

• Focus of this paper is fixed destination multi-depot multiple traveling salesman (robots) problem for indoor networks.

• A methodology is proposed to optimize the task allocation and route planning for building service robots.

• Results are compared with outdoor network based algorithms such as genetic algorithm (GA) and k-means.

• Results emphasize the need for indoor network based algorithms.

• Proposed methodology can be generically applied to several building service robotic applications.

Abstract

Recent advancements in sensing and robotic technologies facilitate the use of on-demand building service robots in the built environment. Multi-robot based systems have arguably more advantages when compared to fixed sensor-based and single-robot based systems. These task-oriented building service robots face several challenges, such as task-allocation and route-planning. Previous studies adopted approaches from other domains, such as outdoor logistics, and made application-specific assumptions. This study proposes a new methodology to optimize the task-allocation and route-planning for multiple indoor robots with multiple starts and destination depots where each robot begins and ends at the same depot (referred to as a fixed destination multi-depot multiple traveling salesman problem-fMmTSP). The performance of the proposed algorithm was compared with two existing outdoor-based algorithms. Results show that the proposed algorithm performs better in almost all the cases for the assumed network, which supports the need to develop algorithms specifically for indoor networks.

Introduction

The operation and maintenance (O&M) phase of buildings contributes towards a majority of their total life cycle costs and environmental impacts. This is primarily because buildings are responsible for a significant portion of the total energy consumption across the globe. For example, buildings are responsible for about 80% of the total electrical energy consumption in the United Arab Emirates, 47% of total energy consumption in China, and 40% of the total energy consumption in the United States [[1], [2], [3]]. In addition, the forecasts suggest that these percentage shares are expected to rise in the upcoming years [4]. However, if ideally operated and maintained, the overall lifecycle cost and environmental impacts of buildings can be significantly reduced [5].

Facility managers employ several techniques to optimally maintain and operate the existing built infrastructure [6]. A few examples include real-time data collection for retrofit decision making, generation and calibration of digital models, identification of energy wastage, delivery of goods and services, periodic maintenance checkups, occupant comfort monitoring, and users and occupants feedback collection. Most of the existing tools to achieve these applications mentioned above require dense instrumentation of the physical space with sensor networks, needs repetitive mundane tasks, and involve a significant amount of labor, time, and resources. Previous studies suggested the use of on-demand automation to address these issues and proposed the deployment of semi and fully autonomous multi robotic systems for indoor building environments [7,8]. These are referred to as building service robots (BSR) and are on the rise due to the computing advancements and evolution in robot cognition, manipulation, modeling, and the ability to interact with humans.

For example, Mudrova and Haves [9] suggest that BSRs can assist with time-dependent tasks such as delivering mail, assisting passengers in airports, guiding customers/visitors in museums/offices for visits/meetings. In addition, their study also developed and tested the deployment of multiple robots for safety, care, and security-related applications such as guided navigation for elderly individuals, periodic monitoring of emergency exits for obstructions, creation of 3D maps of specific rooms at scheduled times, monitored the locations of fire extinguishers, and explored for objects on desks during different times of the day. Researchers are also extending the usage of mobile robots for obtaining customer feedback in hospitals [10]. Similar advantages and applications were suggested by Gao et al. [11] and referred to these robots as patrol service robots. Another study conducted by Chen et al. [12] shows that multi-robot based olfaction (i.e., real-time detection of hazardous gas leakage) could potentially save lives and has more advantages when compared to the state-of-the-art stationary sensor network-based methods. This is arguable because the robot-based method is a more active approach, has the potential to precisely locate the leakage, and is less prone to sensing-related errors. Also, a multi-robot based method is preferred over a single-robot system because of the increased success rate and improved search efficiency. Similarly, studies also suggest that multiple robots (known as a swarm of robots) are more advantageous especially in case of exploration and search and rescue operations because of mutual coordination and collective behavior [[13], [14], [15], [16]].