Actuator Control via Virtual Object Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current safety-related processes using artificial intelligence lack comprehensive monitoring and validation, leading to increased costs and complexity due to the need for redundant systems, which are analytically incomprehensible and thus require extensive checking to ensure safety.
Innovation Solution
A method involving a computing unit and a second computing unit for detecting objects, determining their properties, and comparing them with references to ensure accurate object recognition, allowing secure actuator control without redundant hardware or software, by providing additional objects for enhanced verification and reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If redundant artificial intelligence systems are used to ensure safety, then reliability is improved, but device complexity increases
Solution Approach 1:
The patent uses a second computing unit to create a virtual copy of the object recognition process. This copy detects the same objects and determines the same properties as the first computing unit, allowing for verification without requiring redundant hardware systems. The virtual copying approach ensures safety while avoiding the complexity of multiple physical AI systems.
Solution Approach 2:
The patent introduces a verification mechanism that acts as an intermediary between object detection and actuator control. The second computing unit serves as a mediator that independently verifies the correctness of property determination by the first computing unit, ensuring safety without requiring full redundancy of the AI system.
2Reliability
If comprehensive monitoring and validation of object recognition are implemented, then reliability is improved, but device complexity increases
Solution Approach 1:
The second computing unit creates a virtual copy of the object recognition process, detecting objects and determining properties independently. This allows comprehensive monitoring and validation of the first computing unit's work without requiring complex redundant software systems, as the verification is performed through a parallel but simpler computational process.
Solution Approach 2:
The patent implements a feedback mechanism where the second computing unit's verification results are used to validate the first computing unit's property determination. This feedback loop ensures reliability by continuously monitoring and validating the object recognition process without requiring complex validation software, as the feedback is based on independent but simpler computational verification.
3Measurement precision
If the number of properties to be verified is increased, then measurement precision is improved, but loss of time increases
Solution Approach 1:
The second computing unit performs preliminary verification of object properties in parallel with the first computing unit's detection process. By preparing verification data and performing checks concurrently rather than sequentially, the system can verify multiple properties with high precision without significantly increasing the total verification time, as the preliminary action occurs during the main detection process.
Solution Approach 2:
The patent verifies a selected number of properties rather than all possible properties, applying partial verification where sufficient for safety. The second computing unit determines a subset of properties that are critical for safety verification, avoiding the time cost of verifying every possible property while still achieving high measurement precision for the critical parameters.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
The invention relates to a method (100) for controlling an actuator (201) of a device with a computing unit (203), comprising: - detecting (101) an object (O1, O2, O3) and determining properties (E1', E2', E3') of the object (O1, O2, O3) by a second computing unit (205) based on sensor signals from at least one sensor unit (207); - comparing (103) at least one property (P1', P2', P3') of the properties (E1', E2', E3') of the object (O1, O2, O3) determined by the second computing unit (205) with a corresponding reference (P) by the computing unit (203);and - Control (105) of the actuator (201) by the arithmetic unit (203) taking into account the properties (E1', E2', E3') of the object (O1, O2, O3) determined by the second arithmetic unit (205), if at least one property (P1', P2', P3') of the properties (E1', E2', E3') determined by the second arithmetic unit (205) has a predefined similarity to the reference (P).