Cleaning Robots in Public Spaces: A Survey and Proposal for Benchmarking Based on Stakeholders Interviews

Raphael Memmesheimer, Martina Overbeck, Björn Kral, Lea Steffen, Sven Behnke, Martin Gersch, Arne RoennauRoboCup 2024 · DOI · OpenReview

TL;DR

This paper addresses the need for standardized benchmarking of autonomous cleaning robots in public spaces, a field lagging behind indoor counterparts. By conducting stakeholder interviews and analyzing existing technologies, the authors propose a practical park cleaning benchmark scenario to drive progress and ensure real-world applicability.

What Problem Does This Paper Address?

While autonomous cleaning robots have seen significant development for indoor environments, their application in public spaces remains less explored. This gap presents a challenge for addressing societal needs like labor shortages and maintaining public cleanliness. Existing research and development in this area lack standardized evaluation methods, hindering progress and the creation of robust, adaptable systems. The authors aim to bridge this gap by identifying key requirements from stakeholders and proposing a benchmark that reflects real-world complexities.

Methods

The paper employs a two-pronged approach:

  • Stakeholder Interviews: Seven semi-structured qualitative interviews were conducted with experts from municipal and urban cleaning services, as well as facility management. These interviews aimed to gather insights into current needs, challenges, and expectations for robotic cleaning in public areas. Thematic areas covered included cleaning tasks, technical and economic requirements, acceptance, and market trends.
  • Benchmarking Scenario Design: Based on the interview findings and an analysis of existing cleaning robot technologies, the authors propose a specific benchmarking scenario. This scenario focuses on "Park Pathway Cleaning," a task designed to be representative of real-world challenges and to test various technological capabilities.

Datasets

Not discussed in the paper. The paper focuses on qualitative data from interviews and proposes a new benchmarking scenario rather than using existing datasets for training or evaluation.

Key Insights and Ideas

  • Stakeholder-Driven Requirements are Crucial: The paper emphasizes that successful public space cleaning robots must be designed with direct input from those who will use or be affected by them. This includes understanding their practical needs and concerns.
  • Public Space Cleaning is Inherently Complex: Unlike controlled indoor environments, public spaces present a multitude of unpredictable variables, including diverse debris types, changing weather, dynamic lighting, and the presence of humans and animals.
  • Benchmarking is Key to Progress: The authors draw parallels with other robotics fields (e.g., autonomous driving, search and rescue) where competitive benchmarking has accelerated development and led to more practical solutions.
  • User Acceptance is a Major Hurdle: Beyond technical capabilities, the paper highlights that the acceptance of cleaning robots by both cleaning staff and the public, as well as concerns about vandalism, are significant boundary conditions for their deployment.
  • A Practical Benchmark is Needed: The proposed "Park Pathway Cleaning" benchmark aims to simulate realistic challenges, including varied terrain, debris, and obstacles, to push the development of robust and adaptable cleaning robots.

Key Results

The paper's primary "results" are the insights gained from stakeholder interviews and the proposed benchmarking scenario.

  • Identified Key Cleaning Tasks: Stakeholders identified a range of critical tasks, including collecting small waste (100%), reaching difficult areas (100%), vacuuming (100%), sweeping (86%), and leaf removal (71%).
  • Prioritized Objectives: The most relevant objectives for robotic cleaning were improved cleaning quality (86%), cleaning areas with limited staff capacity (71%), and reducing downtime (71%).
  • Critical Boundary Conditions: The most frequently mentioned boundary conditions included dealing with non-acceptance and vandalism (100%), heterogeneous surfaces (86%), and changing weather/lighting (86%).
  • Proposed Benchmark Scenario: A "Park Pathway Cleaning" benchmark is detailed, specifying cleaning areas (pathway and adjacent wayside), types of contamination (glass, plastic, paper, etc.), and obstacles (curbsides, barriers, cars). Scoring would consider collected item weight/amount and time taken.
  • Technical Requirements: The benchmark is designed to be platform-independent, with technical requirements including a weight limit of 150 kg, size constraints, an emergency button, and robust obstacle avoidance.

Limitations and Future Work

  • Limited Interview Sample: While seven interviews were conducted for public outdoor cleaning, the authors acknowledge this is a relatively small sample size.
  • Lack of Existing Benchmarks: The paper highlights the absence of established, comprehensive benchmarks for public space cleaning robots, necessitating the creation of a new one.
  • Future Work: The authors' immediate future work involves the development and implementation of the proposed benchmarking scenario. Further research could explore more diverse public space environments, advanced sensor fusion for improved perception, and more sophisticated AI for adaptive cleaning strategies. The paper also implicitly suggests the need for research into human-robot interaction and public acceptance strategies.

Summary updated

Abstract

Autonomous cleaning robots for public spaces have potential for addressing current societal challenges, such as labor shortages and cleanliness in public spaces. Other application domains like autonomous driving, bin picking, or search and rescue have shown that benchmarking platforms and approaches in competitive settings can advance their respective research fields, resulting in more applicable systems under real-world conditions. For this paper, we analyzed seven semi-structured, qualitative stakeholder interviews about outdoor cleaning, identified current needs as well as limitations, and considered those results for the development of a benchmarking scenario based on the previous observations.
BibTeX
@inproceedings{memmesheimer2024cleaning,
  author = {Raphael Memmesheimer and Martina Overbeck and Björn Kral and Lea Steffen and Sven Behnke and Martin Gersch and Arne Roennau},
  title = {{Cleaning Robots in Public Spaces: A Survey and Proposal for Benchmarking Based on Stakeholders Interviews}},
  booktitle = {RoboCup},
  year = {2024},
  doi = {10.1007/978-3-031-85859-8\_32},
}
Figure