Capability-based Frameworks for Industrial Robot Skills: a Survey

TL;DR

This survey tackles the lack of standardized descriptions for robot capabilities in industrial settings by conducting a structured literature review of 210 papers. It argues for a taxonomy centered on task, skill, and primitives, showing that most research converges on this hierarchy and on pick-and-place as the primary capable operation, with safety and simple, parameterized approaches dominating industrial contexts. The study analyzes nomenclature, implementation frameworks, and industrial versus non-industrial usage, revealing a landscape where AML/AutomationML and PLC-like frameworks prevail in industry, while ROS and ontologies are more common in non-industrial research. The findings highlight the need for parametric, hardware-aligned capabilities in industrial applications and call for universal representations and standards to enable cross-vendor, cross-site reuse of skills and primitives, ultimately supporting safer and more adaptable manufacturing in HMLV environments.

Abstract

The research community is puzzled with words like skill, action, atomic unit and others when describing robots' capabilities. However, for giving the possibility to integrate capabilities in industrial scenarios, a standardization of these descriptions is necessary. This work uses a structured review approach to identify commonalities and differences in the research community of robots' skill frameworks. Through this method, 210 papers were analyzed and three main results were obtained. First, the vast majority of authors agree on a taxonomy based on task, skill and primitive. Second, the most investigated robots' capabilities are pick and place. Third, industrial oriented applications focus more on simple robots' capabilities with fixed parameters while ensuring safety aspects. Therefore, this work emphasizes that a taxonomy based on task, skill and primitives should be used by future works to align with existing literature. Moreover, further research is needed in the industrial domain for parametric robots' capabilities while ensuring safety.

Figures (4)

• Figure 1: Architecture of the capability-based framework used in this review work to conduct the systematic literature review. The figure shows the hierarchy and the relations between the expressions. The expressions were obtained analyzing the literature on PPR and the domain of robotics research.
• Figure 2: Example of a transmission gearbox assembly. On the left, the general gearbox assembly Process is shown, which is not parameterized and resource independent. The process is composed by the Skills Pick and Place, Pick and Insert, and Pick and Screw. To the right, a specialized task for a particular transmission gearbox assembly is shown. The task is created by matching resources to a process and specifying parameter values for Skills and Primitives. In this case, an object centered representation is used andersen2014definitionhuckaby2012taxonomicji2021learning, the Skill parameter represents the digital artifact of a physical object which contains object specific information (i.e., position). Shaft and housing are passed as parameters to Pick and Place. Gear with bearing is passed to Pick and Insert and Screw to Pick and screw. The artifact's properties can be used to assign parameters also to the underlying Primitives, for example passing an object's target position to a move.
• Figure 3: Wordcloud representing the occurrence of words in the classification table. On the top the names given for task (A1), skill (B1) and primitive (C1). On the bottom the most referred tasks (A2), skills (B2) and primitives (C2). The most common task was assembly, the most common skills pick and place and the most common primitives motion_primitive and open_gripper.
• Figure 4: Diagram showing the differences in capability-based skill frameworks between works with industrial scope and non-industrial scope. The leaves on the right hand side are the most frequently appearing words.