Radar Sensor Blockage Detection Using ML Detection Features
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Automotive radar systems face challenges in distinguishing sensor blockage from the absence of detectable objects, leading to unreliable performance, particularly in environments with few or no large objects in the field of view.
Innovation Solution
A method using supervised machine learning to select and combine features from radar sensor data, such as number of detections, total detection energy, and average detection range, to accurately determine sensor blockage through a logistic regression model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional blockage detection methods (comparing point cloud data between consecutive frames) are used, then the detection process is simple, but the detection accuracy is insufficient and cannot reliably distinguish between stationary objects and actual blockages
Solution Approach 1:
The patent introduces machine learning models as an intermediary between raw point cloud data and blockage detection results. The processing pipeline uses multiple models (semantic segmentation model, instance segmentation model, and detection model) that act as mediators to transform raw sensor data into accurate blockage detections, thereby resolving the contradiction between detection accuracy and system complexity.
Solution Approach 2:
The patent segments the blockage detection task into multiple specialized sub-tasks handled by different machine learning models: semantic segmentation for classifying object types, instance segmentation for identifying individual objects, and detection for determining blockage status. This segmentation allows each model to specialize in a specific aspect, improving overall detection accuracy while maintaining manageable system complexity through modular architecture.
2Measurement precision
If machine learning models are introduced to improve blockage detection accuracy, then detection precision improves, but computational resources and processing time increase
Solution Approach 1:
The patent performs preliminary processing of point cloud data using semantic segmentation and instance segmentation models before the final blockage detection. By pre-processing and organizing the data into structured formats with identified object instances and categories, the system reduces the computational burden on the final detection model, thereby lowering overall energy consumption while maintaining high detection accuracy.
3Reliability
If multiple machine learning models are used for comprehensive blockage detection, then detection reliability improves, but device complexity and processing time increase
Solution Approach 1:
The patent divides the detection process into parallel specialized models that process different aspects of the data simultaneously. The semantic segmentation model processes spatial information, the instance segmentation model processes object identification, and the detection model processes blockage determination. This segmentation into parallel processes improves reliability through comprehensive analysis while managing processing time through concurrent execution.
Solution Approach 2:
The patent implements a feedback mechanism where the detection model receives refined input from the instance segmentation model, which in turn receives input from the semantic segmentation model. Each model's output feeds into the next stage, creating a feedback loop that progressively refines the detection accuracy. This multi-stage feedback approach ensures reliable detection while allowing each stage to build upon previous results efficiently.
Data Source
Figure 1A
Figure 1B
Figure 2
AI summary
A radar sensor includes a memory storing a model defining a relationship between a condition of the radar sensor and a plurality of features of radar detections, the model being generated by a machine learning approach and storing values of the plurality of features associated with the known states of the condition of the radar sensor. A radar detector transmits radar signals into a region, detects reflected returning radar signals from the region, and converts the reflected returning radar signals into digital data signals. A processor receives the digital data signals and processes the digital data signals to generate actual radar detections, each characterized by a plurality of the features of radar detections. The processor applies values of the features of the actual radar detections to the model to determine the state of the condition of the radar sensor.