Semantic Object Marking for Flexible Autonomous Assembly Tasks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automation programming methods are inflexible and machine-specific, making it difficult to manage complex assembly tasks and adapt to changing production processes, as they rely on fixed instructions and do not consider the effect of automated devices on workpieces, limiting scalability and error handling.
Innovation Solution
An autonomous system is developed with a world model that includes data objects representing physical objects, semantic markers to define skills as abstract functions, and an anticipation function for reasoning and error handling, allowing devices to perform tasks based on the environment and available resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional machine-centric programming methods are used, then devices can perform specific assembly tasks, but the system lacks flexibility and scalability when production processes need to be reconfigured
Solution Approach 1:
The patent segments the assembly program into hierarchical levels: assembly objectives (what to assemble), assembly skills (how to assemble), and device behaviors (specific motions). This segmentation allows each level to be independently managed and reused across different production scenarios, reducing overall system complexity while improving adaptability.
Solution Approach 2:
The patent creates universal assembly skills that can be applied across multiple assembly tasks and devices. Instead of writing specific instructions for each machine-part combination, the system defines reusable skills (e.g., pick, place, assemble) that work universally, enabling rapid reconfiguration of production processes without increasing management complexity.
2Reliability
If detailed fixed instructions are written for each assembly task, then specific assembly operations can be performed, but error detection and correction become difficult in long instruction sequences
Solution Approach 1:
The patent implements feedback mechanisms where the system continuously monitors whether assembly objectives are achieved and whether skills are executed correctly. This real-time feedback enables automatic error detection and correction, making reliability high while keeping error detection simple through structured monitoring points in the hierarchical program structure.
Solution Approach 2:
The patent performs preliminary validation of assembly conditions and parameters before executing assembly skills. By checking prerequisites and validating inputs in advance, the system prevents errors before they occur and makes debugging easier by identifying issues at the planning stage rather than during execution of long instruction sequences.
3Adaptability or versatility
If machine-specific programming is used, then current devices can be programmed for specific parts, but the program cannot be scaled to allow other machines to participate without modification
Solution Approach 1:
The patent defines universal assembly skills and standardized data object interfaces that work across different device types and manufacturers. This universality allows programs to be scaled to multiple machines without modification, as each device implements the same skill interfaces and communicates through standardized data objects representing assembly objectives and skills.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
An autonomous system used for a production process includes a device configured to manipulate workpieces according to production process tasks. A device controller generates world model of the autonomous system to include data objects representing respective physical objects in the production process, such as workspace, workpieces, and the device. Semantic markers attached to the data objects include information related to a skill to accomplish a task objective. Semantic markers may be activated or deactivated depending on whether the physical object is currently available for a task performance. The device is controlled to perform tasks guided by the semantic markers while relying on an anticipation function with reasoning operations based on types of physical objects, types of skills, and configuration of the data objects.