Autonomous Vehicle Sensor Fusion for Probabilistic Object Attributes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in accurately predicting object behavior due to limitations in sensor data fusion, where cameras struggle to distinguish between objects based on depth information and LiDAR/Radar are limited in object type detection, leading to inconsistent object classification.
Innovation Solution
A system that processes observation probability distribution functions from multiple sensors to determine attribute presence and adjust vehicle operations based on probabilistic fusion of historical data, using Bayesian filters to generate posterior probability distributions and adjust driving operations accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If cameras are used to detect object types, then object type detection capability is improved, but depth information accuracy deteriorates
Solution Approach 1:
The patent combines data from multiple sensor types (cameras, LiDAR, radar) to create a unified perception system. The camera provides object type detection while LiDAR and radar provide depth information, and the system fuses these data sources to achieve both accurate object classification and depth measurement simultaneously.
2Measurement precision
If multiple sensors are fused to improve detection accuracy, then measurement precision is improved, but system complexity deteriorates
Solution Approach 1:
The patent segments the sensor fusion process into distinct modules: individual sensor data processing, feature extraction, data association, and fusion. This modular approach manages system complexity by breaking down the complex fusion task into manageable components while maintaining high detection accuracy.
3Measurement precision
If sensor data fusion is implemented to distinguish objects, then object classification accuracy is improved, but computational requirements deteriorates
Solution Approach 1:
The system applies partial processing to sensor data by focusing computational resources on the most relevant features and objects. Rather than processing all sensor data in full detail, the system selectively processes data that contributes most to object classification, reducing overall computational energy consumption while maintaining accuracy.
Data Source
AI summary
This document discloses system, method, and computer program product embodiments for controlling a vehicle. For example, the method includes: receiving an observation probability distribution function associated with a target object that was detected by sensor(s) of an autonomous vehicle (AV); identifying a target attribute associated with the target object; detecting a target attribute value associated with the target attribute; and issuing vehicle control instruction(s) that cause AV to adjust driving operation(s) using a future behavior of the target object predicted based on an attribute probability distribution function that defines a probability that the target attribute is actually present for the target object based on probability distribution function(s), wherein the attribute probability distribution function comprises: a probability value associated with the target attribute being present for the target object; and a probability value associated with the target attribute not being present for the target object.