Auditory Sensor Fusion for Hidden Obstacle Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in effectively detecting and avoiding obstacles, particularly those not in the direct line of sight of imaging sensors, such as parked vehicles with their engines running, due to limitations in current sensor systems and algorithms.
Innovation Solution
A system that utilizes a machine learning model trained with both visual and auditory data to identify obstacles, including vehicles with their engines running, by simulating scenarios and integrating sensor data from various sources like cameras, microphones, LIDAR, and RADAR to enhance obstacle detection and collision avoidance capabilities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If imaging sensors (cameras, LIDAR, RADAR) are used for obstacle detection, then the system can detect obstacles in the direct line of sight, but it fails to detect obstacles not in the direct line of sight (e.g., parked vehicles with engines running)
Solution Approach 1:
The patent combines multiple sensor types (imaging sensors, microphones, LIDAR, RADAR) into an integrated sensor system. The machine learning model processes data from all these sensors simultaneously, merging their outputs to achieve comprehensive obstacle detection that overcomes the limitations of individual sensors, particularly detecting both visible obstacles and those hidden from direct view through auditory cues.
Solution Approach 2:
The machine learning model is designed to perform multiple detection functions simultaneously - it processes visual data from cameras and LIDAR, auditory data from microphones, and data from RADAR and other sensors. This multi-functional approach allows the system to detect various types of obstacles using appropriate sensor modalities, with auditory sensors specifically enabling detection of parked vehicles with engines running that are not visible to cameras.
2Reliability
If only visual sensors are used, then the system structure remains simple, but the ability to detect hidden obstacles is insufficient
Solution Approach 1:
The patent merges multiple sensor systems (visual, auditory, LIDAR, RADAR) into a unified detection framework. The machine learning model integrates data from all these diverse sensors, combining their respective strengths to achieve high reliability in collision avoidance while managing the complexity through unified processing architecture.
Solution Approach 2:
The machine learning model serves as an intermediary that processes and integrates data from multiple sensor types. It receives raw data from cameras, microphones, LIDAR, and RADAR, then synthesizes this information to produce reliable obstacle detection results, mediating between the complex sensor inputs and the navigation system's decision-making processes.
3Measurement precision
If a machine learning model processes multiple sensor types, then obstacle detection comprehensiveness improves, but computational requirements and processing time increase
Solution Approach 1:
The system performs preliminary processing of sensor data before full machine learning analysis. By pre-processing auditory data from microphones and visual data from various sensors, the system prepares the data in advance, reducing the computational burden during critical real-time detection phases and enabling faster processing of multiple sensor types without sacrificing detection accuracy.
Data Source
AI summary
A machine learning model is trained by defining a scenario including models of vehicles and a typical driving environment. A model of a subject vehicle is added to the scenario and sensor locations are defined on the subject vehicle. A perception of the scenario by sensors at the sensor locations is simulated. The scenario further includes a model of a parked vehicle with its engine running. The location of the parked vehicle and the simulated outputs of the sensors perceiving the scenario are input to a machine learning algorithm that trains a model to detect the location of the parked vehicle based on the sensor outputs. A vehicle controller then incorporates the machine learning model and estimates the presence and/or location of a parked vehicle with its engine running based on actual sensor outputs input to the machine learning model.