Autonomous Vehicle Gradient Detection via Sensor Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in accurately determining road gradients, leading to incorrect route generation and potential collisions with obstacles due to unstable postures and misinterpretation of road conditions.
Innovation Solution
The method involves using a camera and sensors to capture images, process them into bird's eye views, and calculate gradients by detecting vanishing points to accurately assess road conditions and adjust the driving route accordingly, incorporating a processor to control the vehicle based on these calculations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional camera-based methods are used to estimate road gradient, then the system structure is simple, but the measurement precision of road gradient is low leading to incorrect route generation
Solution Approach 1:
The patent combines multiple sensing systems (camera, laser radar, inertial measurement unit) into an integrated road gradient detection system. The camera captures visual information, laser radar measures distance and surface topology, and the inertial measurement unit provides acceleration data. These multiple sources are fused through coordinate transformation and calculation algorithms to achieve high-precision road gradient measurement that overcomes the limitations of single-sensor approaches.
Solution Approach 2:
The patent introduces an intermediary coordinate transformation process that converts data from different sensor coordinate systems into a unified reference frame. This intermediary step enables accurate integration of camera images, laser radar point clouds, and inertial measurement data, allowing the system to calculate road gradient with high precision by mediating between different measurement modalities.
2Measurement precision
If the vehicle posture is unstable during driving, then the camera-based gradient calculation becomes inaccurate, but adding more stabilization hardware increases system complexity
Solution Approach 1:
The patent implements a feedback mechanism where the inertial measurement unit continuously monitors vehicle posture (acceleration, angular velocity) and provides real-time correction data. This feedback is used to compensate for posture variations in the camera and laser radar measurements, allowing the system to maintain accurate gradient calculations even when the vehicle is moving or on uneven terrain without requiring complex mechanical stabilization.
Solution Approach 2:
The patent replaces mechanical stabilization systems with computational methods. Instead of using gimbals or other mechanical devices to physically stabilize the sensors, the system uses algorithmic compensation based on inertial measurement data to correct for vehicle motion effects in post-processing, significantly reducing mechanical complexity while maintaining measurement accuracy.
3Ease of manufacture
If vanishing point detection is used to calculate gradient, then the method is simple, but it fails on curved roads leading to measurement errors
Solution Approach 1:
The patent creates a universal gradient detection system that can handle both straight and curved roads through multi-functionality. The system uses laser radar to generate point clouds that can represent any road geometry, and applies adaptive fitting algorithms that automatically adjust to detect road characteristics whether straight or curved. This universal approach replaces the vanishing point method's limitation with a flexible solution that maintains reliability across diverse road conditions.
Solution Approach 2:
The patent changes the fundamental parameters used for gradient calculation from vanishing point geometry to laser radar point cloud spatial distribution. By measuring the actual three-dimensional positions of road surface points and calculating gradient from these direct spatial measurements, the system adapts its calculation parameters to work reliably on curved roads where vanishing point methods fail, while maintaining computational efficiency through optimized point cloud processing algorithms.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method to generate an autonomous driving route of an autonomous vehicle includes: receiving an input image of a front view captured from the autonomous vehicle; calculating a gradient between the autonomous vehicle and a ground based on the input image; and generating the autonomous driving route based on the calculated gradient.