Autonomous Vehicle Collision Avoidance Using Audio Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in detecting obstacles that are outside the field of view of imaging sensors, such as parked vehicles with their engines running, which can pose hazards but are not detected due to occlusion.
Innovation Solution
The system utilizes audio data from microphones to identify potential obstacles by processing audio streams through pre-processing modules, machine learning models, and correlation with image and map data to increase confidence in obstacle detection, enabling collision avoidance even when obstacles are not visible to cameras.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If imaging sensors (cameras, LIDAR, RADAR) are used to detect obstacles, then the system can identify visible obstacles and navigate to destination, but obstacles that are occluded or outside the field of view (such as parked vehicles with engines running) cannot be detected
Solution Approach 1:
The system segments the detection task by separating visual detection (imaging sensors) from auditory detection (microphones). Each sensor type handles different aspects of obstacle detection - imaging sensors detect visible obstacles while microphones detect occluded obstacles through sound, particularly engine sounds from parked vehicles. This segmentation allows the system to overcome the limitations of single-sensor approaches.
Solution Approach 2:
Audio data serves as an intermediary to detect obstacles that are not directly visible to imaging sensors. The system uses sound waves as a mediator to gather information about occluded obstacles, particularly using engine sounds as indicators of potentially hazardous parked vehicles. This intermediary approach allows indirect detection of obstacles that would otherwise be completely hidden from visual sensors.
2Reliability
If the system uses multiple sensor types (imaging and audio) to improve obstacle detection, then detection coverage is enhanced, but the system complexity and processing requirements increase
Solution Approach 1:
The system merges imaging sensor data and audio sensor data into a unified obstacle detection framework. The audio data processing pipeline (including pre-processing, machine learning classification, and confidence scoring) is integrated with the existing imaging-based navigation system. This merging allows the system to leverage complementary information from different sensor modalities while maintaining a cohesive decision-making architecture.
Solution Approach 2:
The audio processing system is designed to perform multiple functions: detecting parked vehicles with engines running, identifying occluded obstacles, and providing confidence scores that can be integrated with other sensor data. The machine learning model is trained to classify various audio scenarios and can adapt to different environmental conditions, making the audio subsystem a versatile component that enhances overall system capability without requiring separate specialized systems for each function.
Data Source
AI summary
A controller for an autonomous vehicle receives audio signals from one or more microphones. The audio signals are input to a machine learning model that classifies the source of the audio features. For example, features may be classified as originating from a vehicle. A direction to a source of the audio features is determined based on relative delays of the audio features in signals from multiple microphones. Where audio features are classified with an above-threshold confidence as originating from a vehicle, collision avoidance is performed with respect to the direction to the source of the audio features. The direction to the source of the audio features may be correlated with vehicle images and/or map data to increase a confidence score that the source of the audio features is a parked vehicle with its engine running. Collision avoidance may then be performed with potential paths of the parked vehicle.