Sensor Data Quality Scoring for Autonomous Agent Decisions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous agents rely on sensor data for decision-making, but existing systems lack effective methods to assess data quality and anomalies in real-time, leading to potential inaccuracies and unsafe operations.
Innovation Solution
A data quality and learning system with an edge-based quality analytics engine that employs machine learning and active learning to generate data quality scores, detect anomalies, and assign reputation scores, using processor circuitry to process sensor data from autonomous agents and adjust their operations based on aggregated and fused data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If autonomous agents use sensor data for decision-making, then operational capability is improved, but data quality and reliability deteriorate due to lack of effective quality assessment
Solution Approach 1:
The patent introduces an edge-based quality analytics engine as an intermediary component that sits between the sensor data collection and the autonomous agent decision-making processes. This engine assesses data quality metrics, detects anomalies, and generates reputation scores for different data sources, thereby mediating the reliability issue without preventing the operational capability. The analytics engine processes sensor data and telemetry information to provide quality assessments that guide the autonomous agents in making decisions based on reliable data.
2Reliability
If real-time data quality assessment is implemented, then data reliability is improved, but system complexity increases
Solution Approach 1:
The patent segments the data quality assessment functionality into a separate edge-based quality analytics engine, distinct from the autonomous agent decision-making systems. This segmentation allows the complexity of real-time quality assessment, anomaly detection, and reputation scoring to be isolated in a dedicated module. The autonomous agents can then use the simplified quality metrics and reputation scores without directly implementing complex assessment algorithms, thereby maintaining data reliability while managing system complexity through functional separation.
3Measurement precision
If machine learning and active learning are used for data quality assessment, then measurement precision is improved, but computational requirements increase
Solution Approach 1:
The patent implements preliminary action by pre-training machine learning models and active learning algorithms during system deployment or initial operation phases. The edge-based quality analytics engine learns from historical sensor data and telemetry information to establish baseline quality metrics, anomaly patterns, and source reputations before actual autonomous operations begin. This preliminary learning reduces the computational burden during real-time operations, as the models can make faster inferences using pre-established knowledge rather than performing extensive real-time analysis for every data point.
Data Source
AI summary
Methods, apparatus, systems and articles of manufacture (e.g., physical storage media) to perform data quality assessment and learning for automated agents are disclosed. An example apparatus disclosed herein includes processor circuitry to calculate a data quality score for data generated by sensors of an autonomous agent. The processor circuitry also generates a reputation score based on the data quality score and the data generated by the sensors. The reputation score indicates a level of confidence in an accuracy of the data quality score. Usage of the data by an action circuitry of the autonomous agent is controlled based on the data quality score and the reputation score. The data quality score and the reputation score are a first value and a second value, respectively.