Similar industry standards
International safety standards have been developed such as the ISO 1021836ISO 1348237The safety requirements of different robot categories are described and risk assessments are required for these robots. These guidelines are largely focused on reducing potential risks to the robots and their surroundings. These standards don’t address safety issues or hazards that may be posed by the environments in which robots operate.38.
ISO 1021836This standard addresses safety requirements for industrial robots. The safety principles can be applied to robots or industrial robotic manipulators. However, the standard focuses on the design and construction of robots and their installation within facilities. Safety measures include power and force limiting requirements, stopping functions, and power and force limiting requirements that are focused on robots end39. Operating methods in collaborative applications were also introduced. However, they mainly dealt with control and monitoring systems without much attention to the built environment. The robot’s construction is also subject to the hazards identified in the standard.40.
The ISO 1348236is the only standard for personal care robots41The document outlines safety requirements for mobile assistant robots, physical robots, and person-carrier robots in nonindustrial environments. Safety performance standards are required for assessment and certification.42,43. However, the robot’s performance, design, and safety are more important.44. Also, there are not enough test methods to ensure safety.42. Additionally, standards such as ISO 18646 are available.45ISO 23482 and ISO 2348246Define evaluation methods to evaluate the performance of mobile robot service providers47They are primarily focused on assessing robots’ abilities and ignoring any impediments to robots from the surrounding environment.
RIFMEA is a standardised method to draw out the dangers in the robot’s environment. It also proposes a rating system for assessing the hazards. The RIFMEA allows building owners to identify the hazards and take steps to correct them. The RIFMEA complements existing standards by capturing robot hazards from a new perspective. It also examines the interaction of building elements with robots.
Failure mode analysis and effects analysis (FMEA).
Our proposed framework is based upon the FMEA approach.33FMEA is a systematic way to examine a product’s components or processes in order to identify possible failures and their causes. FMEA can be applied to products and/or processes, but this paper would focus on the building in which the service robot will be deployed. FMEA is an inductive reasoning process that identifies and understands the potential failures in each component of a building to reduce or prevent safety accidents.
FMEA assigns numerical values to each identified failure. The metrics used are Severity (S), Occurrence(O) and Detection [D]. The severity metric is the scale or gravity of the consequence of system failure. The severity of the failure will be assigned a higher value if the consequence is more severe. The Occurrence rating measures how likely it is that a failure will occur. The Occurrence rating is based on how often failures occur. The Detection metric is a measure of the qualitative likelihood that the failure will be detected before it happens. This helps prevent the failure from ever happening. The Detection rating is a measure of how difficult it is to detect failure before it occurs.
When carrying out the FMEA, ratings of S, O, D are assigned to each failure and are multiplied to give a Risk Priority Number (RPN), where RPN=S*O*D. The RPNs are then ordered to create a priority list of failures to resolve. Different rating systems can be used for different FMEA efforts. Some use alphabetical and others use numerical systems.
Literature review on existing FMEA frameworks
FMEA was used in existing research to assess the failures and design of buildings or components. In48, Yang et al. We investigated the possibility of diagnosing and predicting faults in heating, ventilation and cooling (HVAC), systems of buildings. Machado et al.49The accessibility of university faculty buildings was assessed, accounting for persons with disabilities (PWD). The work50An adaptable FMEA framework was used to identify, classify and prioritise latent safety hazards in newly constructed buildings for human usage.51To assess the risks and failures of large-scale building projects, I used an FMEA-based approach. There aren’t yet any papers that explain how the FMEA approach could be used to assess building designs in order to ensure safety for robotic deployments.
This paper is not the only work that shares a similar thought path to this one.27And28This highlighted the need for robots to be able to recognize hazards that are not intended by the robots. They assessed the area of operation, evaluated the quality, and recognized the service robots as key stakeholders in the equation. In27, Dogramadzi et al. Dogramadzi et al. proposed a new hazard analysis called the Environmental Survey Hazard Analysis (ESHA). This method classified potential hazards for autonomous mobile robotics into three categories: Environmental Features and Objects, Agents, and Agents. While the method provided a helpful classification framework to analyze potential environmental-related threats, the paper raised concerns over their application being a relatively shallow breadth-first approach. It also highlighted the difficulty in choosing a hazard-classification scheme that would cover all possible non-mission interactions in any robotic application.
FMEA, on the other hand, may provide a more comprehensive framework for better identifying and categorizing the causes of failures. In this sense, the solutions will be more targeted to ensure service robots are deployed safely and effectively. We have chosen to explore the FMEA approach as an alternative hazard assessment approach. The FMEA approach was adapted to better link failures to the robot’s properties and the design of its environment.
Adapting FMEAs to be robot-inclusive
Robot-inclusive design principles were developed to help analyse failures and provide a framework for categorisation. The metrics that measure how much robot safety is considered in the environment’s design are called robot-inclusiveness. Five robot-inclusive design principles were developed from universal design methodologies. They are safety, accessibility and activity.52,53. Safety is the foundational principle that underpins all four other principles. Accessibility refers to maximising robots’ area coverage and providing connectivity and barrier-free access for robots to move about their tasks. Activity aims at facilitating the efficient integration between workspaces of people, goods, and robots. The Observability principle is about improving the spatial environment for robot vision and perception to navigate and complete tasks. Manipulability, which aims to improve the robot’s ability to move and rearrange objects in its environment with greater precision and success using its end-effectors, is the final principle. These principles will provide an advisory structure to improve spatial environments and work areas in order for robotic deployment.
The RIFMEA approach is an evaluative framework that helps to assess the safety and security of the built environment when robots are applied. It identifies and analyses safety hazards. Fig.2A shows how the proposed RIFMEA process works. It also highlights our additions to and changes to the FMEA methodology. The common FMEA approach to robot safety focuses on the classification, analysis, and determination of failure modes in robot components. The RIFMEA approach considers the risks associated with failure of the robot’s components. In the risk and failure identification stage of the RIFMEA process, it is important to consider both the safety of the robot as well as the safety of the building. To provide clarity on the root cause of failures in robot operation, building components are recorded and categorized under the five robot inclusive design principles.
The key step in applying RIFMEA is to break down the building system into individual components. Through Bachman’s referencing works, we have gained an understanding of the existing frameworks that can be used to categorise and group the various systems and components that make up a building.54Rush55Brand56We have developed a categorisation system for the RIFMEA framework. This structure is illustrated by a building systems diagram in Fig.2B.
RIFMEA Framework for robot-centric and buiding centric principles. (A)Workflow diagram for the proposed RIFMEA procedure. (B)Diagram of the building system.
The building is analyzed in zones according to their programs and activities. The building components are broken down into five categories as shown in Fig.2B. These building components can further be subdivided into their various building elements. For example, an architecture interior building component could be further subdivided into its individual building elements like walls, doors, furniture and floors. The physical parameters of the building components can then be used to analyze them. The diagram of the building system shows how the parts interact to better understand the relationship between them and the failures that occurred.
Causes and effects
The failures can be traced back at the component of the robot and to the physical parameters of the building components. Robot components can be divided into locomotion mechanism (body frame), sensors, manipulators, or end-effectors. The physical parameters of a building component can also be classified into different aspects, such as its form, shape, dimensions, materials, or finishing. It is possible to pinpoint the cause of the failure and suggest targeted actions to reduce the likelihood of it happening again. Our focus is on failures that are directly related to the elements and components of buildings. Failures that are attributed to building components are often caused by problems with the physical parameters or lack thereof. These parameters can be adjusted or altered to reduce the failures.
After a thorough analysis of the causes and effects of each failure, Risk Priority Numbers were calculated. This was done after allocating ratings scales for severity, occurrence, and detection. High priority failures were identified and analysed based upon the degree of mismatch between building components, its elements, and physical parameters that affect the safety of robots operation. To reduce or prevent these accidents, specific building design recommendations are made. These suggestions include changes to the material, form, finishing or form. These design suggestions should consider existing building codes, guidelines, regulations that account for human safety as well as ergonomics.
Modified S.O.D rating system
Individual S.O.D rating system were created to be applicable in the context of service robots being deployed in an environment. They examine the building components as the subject. Clear definition of the scales will ensure impartiality in assigning ratings.33. RIFMEA also adapts its rate system from the MIL-STD-1629A military standard, drafted by The United States Department of Defense57. FMEA was first used by the U.S. Army. It provides a solid foundation for complex military systems and is a reliable foundation. FMEA’s universal applicability means that it can be used in many industries, including automotive, aeronautical and nuclear.58. Different FMEA studies regarding robot safety59,60Robot-human interaction, as well61,62To conduct the FMEA, we have also adopted MIL-STD-1629A. See Table2 below for the modified S, O, and D rating system of RIFMEA. All three scales were modified so that a 1-5 scale was used instead of the 1-4 to provide a neutral rating.63.
The modified Severity scale, which is shown in Table2A and includes three entities: robots (humans), robots (humans), and objects within the robots’ working environment. These were all considered when assessing the severity or failure modes. In this case, humans were given priority, followed by robots, and building components. This is because the design and construction of the built environment are first considered for humans before robotics. Based on previous work in machinery risk assessment, the five categories for severe human injuries were developed.64Safety at work65,66,67.
The modified Occurrence scale (Table2B) measures the likelihood of a failure occurring when the robot is operated in similar environmental conditions. The score that indicates the likelihood of failure is between 0 and 1 is calculated by multiplying the number or instances of failure with the number of interactions the robot had with the building element in question (as illustrated in the Eq).(1) below. This probability score corresponds with an evenly distributed Occurrence scale of 1-5. An Occurrence rating score of 5 or higher would mean a high probability score above 0.875, while a low probability score lower than 0.125 would result in an Occurrence ranking of 1.
$$begin{aligned} chi = frac{text {Number of times failure occurred}}{text {Number of interactions with the building element}} end{aligned}$$
(1)
The Detection Rating is a rating that measures the likelihood of detecting a failure before it occurs. The robot’s autonomy, whether it is fully autonomous, semi-autonomous, or teleoperated, determines the rating. Semi-autonomous robots must consider both the detectability of the human operator and the robot’s autonomous detection capabilities. Table2C shows the scoring.
The robot can create hazard maps using the RPN scores and the location tag for each failure mode. This will allow it to plan better robot tasks routes to avoid dangerous locations. If a location’s RPN score is high, it can be restructured to make it more robot-friendly. It is important to prioritize failures that have higher RPN values, which correspond to higher risk.
Spatial adaptability
If spaces are shared by humans and robots, making changes or modifications to the environment to improve robot safety will inevitably impact how humans use and experience the space. Although robots may be considered new stakeholders in the built environment where possible, human safety and the use of the space should always be the priority. It should support common forms of HRI such as coexistence, cooperation, or collaboration.26,68,69. It is not necessary to alter a space to make robot-inclusive.
When it comes time to remodel a space, we recognize that different building components have different levels spatial adaptability. Building adaptability refers the buildings ability to accommodate changes.70. It would be more difficult to modify the design or relocate the columns (Structure), than to simply change the furniture layout (Plan). However, there are different solutions that can be used to solve a single problem. They may have different effectiveness, but they may be more adaptable. To address the danger of unorganized cables on the floor, you could either rewire the cables or use cable management methods such as installing cable trunking. Although the former is more costly and more labor-intensive than the latter it could be more effective in eliminating any hazard. The planning for robot-inclusive spaces will be influenced by the ability to adapt.