Enhancing Worker Safety in Harbors Using Quadruped Robots

Technical Summary: Enhancing Worker Safety in Harbors Using Quadruped Robots

Problem Statement
The harbor environment presents complex challenges for infrastructure inspection, characterized by dynamic operations, diverse terrains, and significant safety risks for human workers. While various robotic solutions exist (wheeled, aerial, and legged), each has limitations in this specific context. Drones offer speed and aerial perspectives but suffer from limited battery life, regulatory restrictions (e.g., flight cones near airports), and safety concerns in confined spaces or near personnel. Wheeled robots lack the mobility required for uneven surfaces and obstacles. Quadruped robots are proposed as a viable alternative due to their agility, ability to navigate stairs and obstacles, longer battery life, and capacity to operate in close proximity to workers without special permits. However, there is a lack of literature addressing the specific application of quadruped robots for land-based harbor inspections, with existing studies focusing primarily on underwater structures or general civil environments.

Methodology
To address the gap between robotic capabilities and actual harbor safety needs, the authors conducted a preliminary study at the Port of Genova, Italy. The methodology involved a series of semi-structured interviews with 10 stakeholders across three distinct terminals: Terminal IMT (containerized and hazardous goods), Terminal C. Steinweg - GMT SRL (aluminum products), and Terminal Rinfuse Genova (bulk goods). The interviewees included safety managers, coordinators, and operational workers.

The interviews followed a three-phase structure:

1. Preliminary Phase: Gathering information on workflows and identifying critical safety areas.
2. Demonstration Phase: Showcasing the motion capabilities of the Spot robot (Boston Dynamics) via teleoperation.
3. Design Phase: Collaboratively envisioning applications for the robot and gathering worker perceptions.

Notably, the researchers withheld details of their prior research until the end of the meetings to ensure participants formed unbiased opinions on potential use cases. The data was transcribed and analyzed to categorize problems by location, priority, and frequency, and to map them to potential robotic solutions and autonomy levels.

Key Contributions
The paper makes two primary contributions:

1. Identification of Critical Areas: It systematically identifies and classifies safety-critical scenarios within the harbor environment, categorizing them by location (Storage Yard, Ships' Hold, Cranes) and priority.
2. Feasibility Assessment of Quadruped Robots: It evaluates which of these identified scenarios are suitable for deployment by a sensorized quadruped robot, contrasting them with drone-based solutions and defining the required level of autonomy (teleoperation, mixed autonomy, or full autonomy).

Results
The analysis of the interviews yielded a taxonomy of 15 specific safety problems, which were mapped to potential robotic interventions (Table I and Table II in the original text).

• Storage Yard (60% of critical areas): This area includes hazardous goods parks and bulk storage. The study found that quadruped robots are highly suitable for monitoring chemical leaks, radiation, fires, and self-combustion in bulk storage (e.g., coal). Their ability to climb over piles and navigate tight spaces allows for precise thermal and gas sensing. They are also preferred for visual inspections of container stacking and train seals, where stability and proximity are crucial.
• Ships' Hold (27%): For inspections inside ship holds (e.g., checking air quality, cargo stability, or radioactive materials), the study suggests a "mixed autonomy" approach. While teleoperation is possible, the crowded and dynamic nature of the hold necessitates autonomous behaviors (exploration, mapping) to handle connection instability and obstacles. Quadruped robots are favored here for their ability to navigate under cargo and over small obstacles, providing a 3D perspective via LiDAR that drones cannot easily achieve in confined vertical spaces.
• Cranes (13%): The study notes that for crane inspections (e.g., X-ray or penetrating liquid application), workers suggested drones as the preferred solution due to the difficulty of reaching elevated heights. Quadruped robots were not deemed the primary solution for these specific aerial tasks.

The results indicate that most high-priority tasks in storage yards and ships' holds require repeated or event-driven inspections. Consequently, fixed-route programming is insufficient; the robots require capabilities for exploration, mapping, and feature detection.

Significance and Claims
The paper positions this work as an initial, foundational phase toward developing a robotic solution for harbor safety. The authors modestly claim that their primary value lies in engaging end-users (workers and safety managers) to tailor robotic solutions to real-world requirements. By involving workers in the design process, the study aims to address concerns regarding job displacement and precision, framing the robot not as a replacement for human labor but as a safety instrument.

The study concludes that while quadruped robots are not a universal solution for all harbor tasks (particularly high-altitude crane inspections), they offer a distinct advantage for ground-level and confined-space inspections where agility, sensor payload capacity, and worker proximity are paramount. This work serves as a basis for future developments aimed at implementing and testing these robotic solutions in the harbor environment.