Robotic systems' flexibility, precision, and effectiveness are changing various industries. As these systems grow more versatile, a complete model of their capabilities becomes increasingly important. The Robotic Capability Ontology (RCO), an application ontology created explicitly for capturing robotic capabilities, is presented in this paper. Modeling robot capabilities according to user manuals or design/technical specifications from manufacturers is the main objective of RCO. The RCO utilizes the Manufacturing Service Description Language (MSDL), a domain reference ontology designed for manufacturing services and aligned with the Basic Formal Ontology (BFO) and Information Artifact Ontology (IAO). MSDL's modular structure and manufacturing domain-neutral classes allow RCO to describe robotic capabilities and expand upon them precisely. For those seeking to install and use robots, RCO has significance because it promotes collaborative behavior, design help, appropriate use of resources, enhancements in efficiency, and optimized planning. Additionally, it encourages research initiatives to develop. The development and applicability of the RCO are covered in greater length in this article.
We are on the leading edge of a robotics revolution that promises to transform the world in
multiple dimensions as the new era of modern technology takes shape. Robotics has wholly
transformed our society and economies [
We must represent the idea of capability in robotics, collecting pre-designed capabilities via
the robot’s manufacturer design specification or robot manual to determine the robots’ factual
capability. An ontological model can offer the framework for recognizing and distinguishing
these capabilities. We demonstrate the design of an ontological model that precisely captures
these notions of the robot from its design specifications. We utilize Manufacturing Service
Description Language ontology[
The remainder of this article is structured into the following three sections; Section 2 provides a brief overview of robotics scholars’ literature studies and how capability modeling has a truly greater impact on the future of robotics. It also discusses the importance of ontologies and how they can help model robotic systems’ capabilities. Section 3 emphasizes reasons why modeling robotic capabilities is critical, especially in applications in industrial Use cases. Section 4 examines the proposed approach for modeling robotic design specifications using (MSDL) and explains how we model robotic capabilities and section 5 validation of the ontology model via SPARQL at last. Section 6 prescribes upon conclusion and how we further expand this approach. Finally, This paper clarifies the significance of ontological approaches for modeling notion capability in robotics.
These enormous advances in robotics are about more than merely automating tasks; it is also
about giving robots abilities that closely resemble, and in some cases meet or exceed human
capabilities [
to us can take. Sabbagh & Ameri defined “Capability” as the inherent potential of diverse resources in generating benefits for various stakeholders [ 21]. While Sarkar & Sormaz argued that capability, capacity, and competency are mostly used with interconnected connotations [22]. Francesco & Stefano talks about what are capabilities and capacities, how they are similar to each other in terms of functionality, and how you differentiate as they state, “capacities essentially indicate how the corresponding capabilities can be practically implemented.” also, they perform different analysis based on using the top-level ontology DOLCE as a framework [23]. So based on these definitions, we can state that capability is ” The ability or potential of an entity, process, or product to accomplish a desired result or execute specific tasks under a set of established parameters.” In robotics, a robot’s capability can relate to performing tasks such as lifting, placing items, navigating areas, detecting objects, or understanding voice. But these are the abilities or tasks a robot is created and programmed from the start to accomplish.
We denfie it as “Pre-designed capability” because what a robot can perform under actual circumstances could vary from its pre-designed capabilities mentioned in the manufacturer’s design specification or robot manual. A robot’s pre-designed capability, which could have been viewed as its fundamental function, could differ from its actual capabilities observed in practical application contexts. While these functions are addressed in the manufacturer’s design specification or robot manual, a robot’s actual capabilities may diverge dramatically from these pre-planned functions. A robot, for example, maybe pre-designed to perform a specific task in a production line, such as component assembly. Still, in practice, it may be repurposed to transport equipment across the manufacturing facility, a capability the designer or manufacturer may not have originally intended. In an entirely different setting, a robot developed for a specific function may even be used as a weight in a gym, which is a different unanticipated outcome.
In this section, we describe various reasons modeling robotic capabilities is critical, especially in applications in industrial Use cases.
• Modeling these capabilities can guide the selection of appropriate robots for specific tasks, leading to improved productivity [24]. • Robots are costly, and a lack of knowledge or modeling of their capabilities can lead to inefficient usage and diminished resources [ 25]. For example, if a robot’s reach or precision is incorrectly represented, it may be assigned tasks that it cannot accomplish adequately, resulting in inefficiency and waste. • Robots can endanger human beings if employed outside their capabilities or for jobs not built for that particular task[26]. For instance, understanding a robot’s lifting capability is crucial to prevent it from being overloaded, which could lead to equipment failure or even harm human operators and other equipment. • Automated Decision-Making towards trustful and flexible manufacturing, promoting reconfiguration in production lines [27].
In the modern era of flexible production systems, robotic capability modeling has grown progressively relevant when robotics is anticipated to carry out various activities [28, 29, 30]. In this context, modeling robotic capabilities helps lead the development of adaptable robotics capable of effortlessly shifting among diverse jobs. Also, modeling robotic capabilities isn’t simply an abstract activity but an important part of the industrial process. It influences robotics’s development, installation, and secure functioning in industries, impacting productivity, trustfulness, cost efficiency, and security.
To design an effective robotic system or a single robot to perform a particular task, operators, engineering, or people in charge must first grasp the required capabilities of a robot. Ontologies are an excellent answer to this problem. Ontologies can accurately explain and represent complex, constantly evolving systems like robots with varying capabilities and capacities. Several recent works in the literature regarding robotics utilized ontologies to model robotics automation for modeling common concepts, robotics standards, and human-robot interaction [31, 32]. A more nuanced knowledge of what is expected from a robot, including minor skills like being able to perform complicated operations or adjust to evolving circumstances, can be provided through an ontology-based approach to robotics capability modeling. For example, consider three robots that have been trained to do certain tasks: Robot-A assembles parts, Robot-B paints ifnal items, and Robot-C bundles them up for delivery, but when production methods evolve and additional products are added to be produced, they change the production order or add the different specification to the product. Robots existing capability information, which depends purely on pre-designed capability, does not make adapting to these new requirements simple. All robots must be reprogrammed for each new duty, which is expensive and time-consuming and causes delays in production and financial losses. However, suppose the robots can adapt to this change. In that case, they may be able to adapt to new tasks more rapidly while still preserving production efficiency and lowering costs.
To do this, we must represent the idea of capability in robotics, collecting pre-designed capabilities via the robot’s manufacturer design specification or robot manual to determine the robot’s factual capability. An ontological model can offer the framework for recognizing and distinguishing these capabilities, enabling researchers to develop and deploy more flexible robots. One important thing to note before discussing the methodology and scope of the Ontology is that RCO has a broader scope in terms of attaining the objectives, which includes (Capability, Function, Quality, and Role) as this work is a part of the project CHAIKMAT 2. But, in this paper, we stick to the notion of capability. We use the NED2 Robot3 and its manufacturer Hardware Design Manual4 as a use-case scenario in the manufacturing industry to model the capability of a robot. We extract some parameters such as Payload = 300 g, Reach = 440m, and Repeatability = 0.5 mm from the technical specifications of NED2 Robot provided by its manufacturer. We use these parameters to model Payload as picking capability, Reach as reach capability, and Repetition as repeatability capability, where the robot’s essential functions are to perform handling and transportation. Besides classifying and modeling different capabilities, our ontology should provide insightful responses to important questions concerning these robot capabilities. These Competency Questions (CQ) are tested in section 5.
CQ1: Can a NED2 robot repetitively perform an operation with a precision of +- 0.5 mm? ;
CQ2: Can a NED2 robot pick up a 300-gram item? ; CQ3: Can a NED2 robot pick up an item 440mmm distant from its base? We made the choice to emphasize the practical value of our work when developing our ontology by using real-world scenario-based competence questions. The events we encountered or imagined in a real production line environment served as the basis for our research. The questions we asked, while particular, are typical of many other comparable problems that can come up when deploying robots for automation. For instance, we picked the robot NED2 for our examples because it represents the general class of robots used in comparable production situations, not because our questions are confined to this specific robot. We intended to illustrate how our ontology may address particular, practical issues that emerge when deploying robots in a production environment by concentrating on actual, concrete examples.
4.0.1. Applying MSDL in Modeling Robot Capabilities Key concepts that we used from MSDL are the following MSDL they defined Capability “A disposition in whose realization some agent has an interest”; Engineerined Artificat conceptualize refers to “An object or object aggregate that is deliberately created to have a certain function and is prescribed by some design specifications.” Another critical factor to using MSDL is there are measurement datum for different measurement labels, which are required to model values of different values such as length, and volume; MSDL has partially aligned BFO as well as Open Biological and Biomedical Ontology Foundry (OBO)5 AND Information Artifact Ontology (IAO)6.
The RCO extends MSDL’s structure as it can expand the collection of classes and relationships by introducing new domain-specific notions required for capturing robot capabilities, as shown in figure 1. We’ve defined the class RCO: HandlingandTransportationEquipment within the MSDL:Equipment class to describe equipment specially built for material handling and transportation. This class is a subclass of MSDL:EngineeredArtifact, which represents physical and digital items produced through engineering design and production techniques. Furthermore, we’ve introduced the class RCO:Robot as a subclass of RCO:HandlingandTransportationEquipment, implying that robots are specific equipment for material handling and transportation. As a result, instances inside our ontology, such as the RCO:NED2Robot, can be defined under the RCO: Robot class, indicating that it is a component of the larger domain RCO:HandlingandTransportationEquipment, and subsequently MSDL:Equipment. We’ve amended the MSDL’s Capability class, a subclass of the BFO:Disposition class of BFO:Entity to depict distinct robot capabilities explicitly. This class relates to an Objective representing an entity’s ability to achieve targets connected with its overall purpose. We added the RCO:MaterialHandlingCapability as a subclass of MSDL:ManufacturingCapability and subsequently, the RCO:RepeatabilityCapability
class as a subclass of MaterialHandlingCapability. The IAO:MeasurementDatum class is a generic class to represent measurement data. IAO:ScalarMeasurementDatum is a subclass of IAO:MeasurementDatum , and IAO:LengthMeasurementDatum is a subclass of IAO:ScalarMeasurementDatum to properly define length-related measurement data. We developed a special instance named RCO:NED2RepeatabilityMeasurement within the IAO:LengthMeasurementDatum class to capture the repeatability measurement value. The object attribute MSDL:hasCapability links entities to their capabilities. Thus, using the MSDL:hasCapability object property, the NED2 instance is connected to the RCO:RepeatabilityCapability.
Moreover, the object attribute RCO:isMeasurementOf is developed to represent the link between measurements and capabilities. IAO:MeasurementDatum , a subclass of BFO:InformationContinuantEntity, is the domain of the RCO:isMeasurementOf, and the range is ’BFO:specificallyIndependentContinuant , representing the measurement’s related specific independent continuant entity. The Basic Formal Ontology (BFO) relationship BFO:hasPart represents the lowest and highest values of the NED2 repeatability capability and transmits measurement values, measurement data, and parameter units. The property IOA:hasmeasurementUnitLabel shows the parameter unit’s label, whereas IOA:hasMeasurementValue associates the measurement value with a specific datum. This structure results in an accurate and comprehensive depiction of robotic capabilities in ontology. This ontological framework supports the exact and exhaustive representation of robotic capabilities, emphasizing the need to use RCO in the context of robotics in the manufacturing Industry. We model payload and Reach as a capacity subclass of Capability because of its parameters of capabilities. For Payload, we use Mass Measurement Datum; for Reach, we use the exact length measurement datum as repeatability.
We combine all 3 CQS into one: ” Can a NED2 robot repetitively operate with a precision of +- 0.5 mm, pick up a 300-gram item 440mm distant from its base?” As a result, we could answer our competence question with a binary ” Yes” or ” No,” indicating if the NED2 robot could fulfill all of the requirements at once. We used the GraphDB SPARQL API to run a single comprehensive SPARQL query, as Shown in Fig. 2, to evaluate our (RCO) Competency Questions. Knowing the precise response is essential, but understanding the query’s details and being able to respond to a variety of comparable requests are also important. Using ontologies and SPARQL queries for robotic capabilities improves knowledge modeling and interoperability while providing scalability and automation assistance, allowing for efficient, data-driven decisionmaking across various systems and applications. Information may be automatically retrieved and analyzed using SPARQL queries, as seen in the NED2 robot example. This can considerably minimize the amount of time and effort required to answer difficult queries concerning a robot’s capabilities. Combined with other systems, ontology and SPARQL investigations can give helpful decision-making information. A manufacturer, for example, can rapidly decide if a given robot ifts a specific task based on its established capabilities. This allows for data-driven decisions, which improves efficiency and results.
Our work demonstrates the significance of ontological models in correctly capturing robotic capabilities to support industrial flexibility. The study underlines the importance of semantically exact models as robots grow capable of performing various functions. An ontological model enables a thorough comprehension of concepts like reach, accuracy, repetition, and various robotic behaviors. The described ontology-based method has several advantages in industrial situations and provides a foundation for modeling robot design criteria. Exploiting ontological models helps select appropriate robots for specific jobs, enhances productivity, reduces inefficient usage and possible dangers, and promotes trustworthy and adaptable production by precisely capturing the capabilities of robots. Overall, using ontological modeling in robotic capability modeling advances flexible production systems and improves the creation and use of autonomous robots. Our ongoing work focuses on modeling notions such as function, quality, and role of robots to enhance the understanding and utilization of robotic capabilities in flexible manufacturing systems.
This work is performed within the CHAIKMAT project funded by the French National Research Agency (ANR) under grant agreement ” ANR-21-CE10-0004-01”.