Autonomous Vehicle Sensor Fusion for Reliable 3D Object Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current autonomous vehicle technologies are not safe nor sufficiently reliable for commercial deployment due to challenges in detecting dynamic driving environments, such as vehicles, pedestrians, and obstacles, while maintaining affordability and sensor efficiency.
Innovation Solution
A computer-based architecture that processes inputs from multiple sensor modalities, including cameras, lidars, and radars, to generate a unified perception of the environment by fusing sensor data, adjusting weight contributions based on sensor performance, and synchronizing asynchronous detection results for improved object detection and localization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple sensor modalities are used to improve detection accuracy, then object detection reliability is improved, but system cost and complexity increase
Solution Approach 1:
The patent combines multiple sensor modalities (camera, lidar, radar) into a unified sensor fusion architecture that processes data from all sensors simultaneously. The system merges detection results from individual sensors through association and fusion algorithms, creating a comprehensive environmental perception that improves reliability while managing complexity through integrated processing.
Solution Approach 2:
The sensor fusion module serves multiple functions: it processes data from different sensor types, performs object detection, tracks objects across frames, and generates unified detection results. This multi-functional approach allows a single system component to handle diverse sensor inputs and produce comprehensive output, reducing overall system complexity.
2Measurement precision
If advanced sensor fusion algorithms are implemented to improve detection accuracy, then measurement precision is improved, but processing time increases
Solution Approach 1:
The system performs preliminary object association and detection before final fusion. Detection results from individual sensors are pre-processed and associated with potential objects in advance, which reduces the computational burden during the final fusion step and enables faster processing while maintaining high precision.
Solution Approach 2:
The sensor fusion system uses self-service mechanisms where detection results from one sensor modality help validate and refine detections from other sensors. The system automatically associates detection results with objects and performs confidence-based filtering, reducing the need for extensive manual processing and improving efficiency.
3Reliability
If high-performance sensors are deployed to improve detection reliability, then object detection reliability is improved, but system cost increases
Solution Approach 1:
The system dynamically adjusts processing parameters and sensor activation based on environmental conditions and detection needs. By changing operational parameters such as sensor sampling rates, processing intensity, and fusion algorithms selected, the system maintains high detection reliability while optimizing resource usage and reducing overall system cost.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enhances the safety and reliability of autonomous vehicles by providing a robust environmental perception system that accurately detects objects and environments, reducing the need for costly sensor upgrades and improving overall system performance.
Implementation Method 1
an image capturing device to generate image data of the scene
Implementation Method 2
an image capturing device to generate image data
Implementation Method 3
a lidar apparatus, including a transmitter configured to emit light comprising light pulses toward a scene including one or more target objects and a receiver configured to detect reflected light
Implementation Method 4
detect reflected light from one or more of the target objects
Data Source
AI summary
A process for sensing a scene. The process includes receiving sensor data from a plurality of sensor modalities, where each sensor modality observes at least a portion of the scene containing at least one of the objects of interest and generates sensor data conveying information on the scene and of the object of interest. The process further includes processing the sensor data from each sensor modality to detect objects of interest and produce a plurality of primary detection results, each detection result being associated with a respective sensor modality. The process also includes fusing sensor data from a first sensor modality with sensor data from a second sensor modality to generate a fused 3D map of the scene, processing the fused 3D map to detect objects of interest and produce secondary detection results and performing object level fusion on the primary and the secondary detection results.