Dynamic Sensor Engagement for AI Inference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In dynamic, noisy, and unpredictable physical environments, existing sensor systems face inefficiencies due to sub-optimal sensing modalities, leading to poor AI task performance and high energy consumption, as noise in sensor data affects downstream AI tasks and energy usage.
Innovation Solution
A method that dynamically engages sensors by processors to obtain raw data, applies an outlier detector to determine if data is usable, and automatically activates additional sensors if necessary to collect new data, using machine learning models to derive inferences for downstream tasks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple sensors of different modalities are engaged continuously to ensure robust data collection, then the reliability and completeness of sensor data is improved, but the energy consumption increases significantly
Solution Approach 1:
The system dynamically adjusts sensor engagement based on real-time conditions. The outlier detector continuously monitors data quality and automatically activates additional sensors only when needed, rather than keeping all sensors active continuously. This dynamic approach ensures reliable data collection when required while minimizing energy consumption during normal operation.
Solution Approach 2:
The system employs self-service mechanisms where the outlier detector automatically identifies when additional sensors are needed and triggers their activation without external intervention. The machine learning models and outlier detection algorithms autonomously manage sensor resources, making decisions about when to engage additional modalities based on data quality assessment.
2Productivity
If sensors are engaged to collect data in dynamic and noisy environments, then the ability to perform AI tasks is improved, but the data quality deteriorates due to noise and unpredictability
Solution Approach 1:
The system implements feedback loops where the outlier detector continuously monitors the quality of collected data and provides feedback to the sensor engagement system. When data quality deteriorates due to noise or environmental factors, the feedback mechanism triggers activation of additional sensors to compensate, ensuring maintaining AI task performance.
Solution Approach 2:
The outlier detector acts as an intermediary between the sensors and the AI processing pipeline. It filters and assesses data quality before data is used for AI tasks, and can trigger additional sensing when the intermediary detects poor quality, thereby protecting the AI system from noisy or unreliable data.
3Adaptability or versatility
If all sensors are activated simultaneously to handle unpredictable environments, then the adaptability to various conditions is improved, but the device complexity and energy usage increase
Solution Approach 1:
The sensor system is segmented into multiple modalities that can be independently controlled. Rather than treating all sensors as a single complex unit, the system divides them into separate sensor groups that can be activated selectively based on environmental conditions and task requirements, reducing overall system complexity while maintaining adaptability.
Solution Approach 2:
The system changes operational parameters dynamically by adjusting which sensor modalities are active based on environmental conditions. The outlier detector monitors data quality parameters and triggers parameter changes in sensor engagement, allowing the system to adapt to unpredictable environments by adjusting sensor activation patterns rather than maintaining fixed complex configurations.
Data Source
AI summary
A method, computer program product, and system include a processor(s) that engages, based on a request for an inference, from a group of sensors of multiple modalities at a physical location, sensor(s) of a main modality to provide data to a pipeline to generate the inference. The pipeline includes one or more machine learning models which generate the inference for a downstream task. The processor(s) obtains raw data from the sensor(s) of the main modality and applies an outlier detector to the raw data. Based on determining that there is an outlier the processor(s) automatically engages sensor(s) of at least one different modality than the main modality from the group of sensors of multiple modalities and obtains new raw data from the sensor(s) of the at least one different modality. The processor(s) applies the one or more machine learning models to the new raw data to derive the inference.