Vision-based Autonomous Disinfection of High Touch Surfaces in Indoor Environments

Sean Roelofs, Benoit Landry, Myra Kurosu Jalil, Adrian Martin, Saisneha Koppaka, Sindy K. Y. Tang, Marco Pavone2021 · DOI · OpenReview

TL;DR

This paper introduces the smallest untethered UAV capable of autonomous indoor disinfection, specifically targeting high-touch surfaces like door handles. By combining vision-based object detection, precise spraying parameters, and autonomous navigation, this system offers a novel approach to mitigating the spread of infectious diseases in indoor environments.

What Problem Does This Paper Address?

The COVID-19 pandemic highlighted the critical role of autonomous systems in public health responses. While Unmanned Aerial Vehicles (UAVs) have been explored for mass disinfection, existing research often focuses on outdoor or large-scale applications, or relies on semi-autonomous systems. There's a significant gap in developing small, fully autonomous UAVs capable of navigating complex indoor spaces and precisely targeting high-touch surfaces, which are known vectors for various pathogens beyond SARS-CoV-2. This paper aims to bridge this gap by creating a UAV that can autonomously identify, approach, and disinfect individual high-touch surfaces in unstructured indoor environments, thereby improving efficiency and reducing human exposure.

Methods

The core of this work is a custom-built UAV designed for autonomous indoor disinfection. Key methodological contributions include:

  • Hardware Integration: A small, untethered quadrotor base (Holybro S500) is equipped with a Nvidia Jetson Xavier NX for onboard computation, Intel RealSense cameras (D435i for depth, T265 for pose estimation) for perception, and a custom 3D-printed mount for sensors. A peristaltic pump and a fine-nozzle sprayer are integrated for disinfectant application.
  • Vision-based Object Detection and Localization: The system utilizes a YOLO object detection model to identify doors and handles. A novel method for handle localization is proposed, which involves projecting a raw centroid from the handle's point cloud onto a fitted planar model of the door to reduce noise and improve accuracy, especially at a distance.
  • Autonomous Navigation and Planning: The UAV builds a 3D occupancy map of its environment using its sensors. For path planning, it employs the RRT* algorithm from OMPL to find collision-free 3D paths. Path following is handled by a cascaded PID controller via the PX4 flight controller.
  • Spraying Parameter Optimization: The paper investigates crucial spraying parameters for effective disinfection. They determined that 70% ethanol is a suitable disinfectant and identified optimal spraying distance (30 cm) and duration (2 seconds) to ensure a persistent film on the target surface for at least 60 seconds, sufficient to neutralize common viruses.
  • State Machine for Autonomous Operation: A high-level state machine orchestrates the UAV's behavior, transitioning between "takeoff," "explore" (searching for handles while mapping and avoiding obstacles), "spray" (approaching and disinfecting a detected handle), and "land" states.

Datasets

Not discussed in the paper. The system relies on real-time perception and mapping of the environment rather than pre-trained datasets for its core functionality.

Key Insights and Ideas

  • The "Smallest" Autonomous Disinfection UAV: The paper claims to have built the smallest untethered UAV with both full autonomy and spraying capabilities, enabling operation in confined indoor spaces. This is a significant engineering feat.
  • Targeted Disinfection is Key: Instead of broad area spraying, the system focuses on individual high-touch surfaces (door handles). This approach is more efficient in disinfectant usage and potentially more effective against fomite transmission.
  • Vision-based Localization in Unstructured Environments: The novel method for handle localization, which projects a centroid onto a planar door model, is crucial for robustly identifying targets at a distance in cluttered indoor settings.
  • Empirical Determination of Spraying Parameters: The careful experimental determination of spraying distance and time, validated by surface coverage tests, is essential for ensuring the disinfectant's effectiveness without excessive waste or risk of collision.

Key Results

  • The UAV successfully and autonomously identified and sprayed door handles in 10 out of 10 trials in a controlled motion capture environment.
  • During spraying, the UAV maintained an average distance of 6.8 cm from the target spray pose, with relatively small deviations within individual trials, indicating good position control.
  • The determined spraying parameters (30 cm distance, 2-second duration) were shown to create adequate coverage (≥90% for 1 min) for neutralizing common viruses with 70% ethanol.
  • The system can build a 3D map of its environment and plan collision-free paths around obstacles.

Limitations and Future Work

  • Disinfectant Capacity: The current UAV can only carry enough disinfectant for approximately 20 door handles, limiting its operational duration.
  • Surface Versatility: The system is currently optimized for door handles and may require modifications to effectively disinfect a wider variety of surfaces.
  • Flight Time: The payload requirements for spraying significantly impact flight time.
  • Future Work: The authors suggest tethered UAVs to overcome payload limitations, allowing for larger disinfectant capacity and longer flight times. They also propose adjustable nozzles for spraying different surfaces and directions, and improved localization strategies for even greater accuracy.

Summary updated

Abstract

Autonomous systems have played an important role in response to the Covid-19 pandemic. Notably, there have been multiple attempts to leverage Unmanned Aerial Vehicles (UAVs) to disinfect surfaces. Although recent research suggests that surface transmission is less significant than airborne transmission in the spread of Covid-19, surfaces and fomites can play, and have played, critical roles in the transmission of Covid-19 and many other viruses, especially in settings such as child daycares, schools, offices, and hospitals. Employing UAVs for mass spray disinfection offers several potential advantages, including high-throughput application of disinfectant, large scale deployment, and the minimization of health risks to sanitation workers. Despite these potential benefits and preliminary usage of UAVs for disinfection, there has been little research into their design and effectiveness. In this work, we present an autonomous UAV capable of effectively disinfecting indoor surfaces. We identify relevant parameters such as disinfectant type and concentration, and application time and distance required of the UAV to disinfect high-touch surfaces such as door handles. Finally, we develop a robotic system that enables the fully autonomous disinfection of door handles in an unstructured and previously unknown environment. To our knowledge, this is the smallest untethered UAV ever built with both full autonomy and spraying capabilities, allowing it to operate in confined indoor settings, and the first autonomous UAV to specifically target high-touch surfaces on an individual basis with spray disinfectant, resulting in more efficient use of disinfectant
BibTeX
@misc{roelofs2021vision,
  author = {Sean Roelofs and Benoit Landry and Myra Kurosu Jalil and Adrian Martin and Saisneha Koppaka and Sindy K. Y. Tang and Marco Pavone},
  title = {{Vision-based Autonomous Disinfection of High Touch Surfaces in Indoor Environments}},
  year = {2021},
  eprint = {2108.11456},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2108.11456},
}
Figure