Vehicle Visual Odometry Using Eccentricity Maps and Key Frames
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current vehicle navigation systems rely on computationally intensive dense optical flow calculations for visual odometry, which are inefficient and time-consuming, limiting the accuracy and speed of determining vehicle motion data in autonomous or semi-autonomous modes.
Innovation Solution
The method involves creating an eccentricity map from video image data, combining it with RGB images, and processing them using a deep neural network to generate vehicle motion data, significantly reducing the number of calculations required by replacing dense optical flow calculations, thereby improving the efficiency of determining vehicle location and orientation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If dense optical flow calculations are used for visual odometry, then measurement precision of vehicle motion data is improved, but productivity is worsened due to computational intensity
Solution Approach 1:
The patent segments the video stream into individual video frames and processes only key frames rather than all frames. This segmentation approach maintains measurement precision by selecting representative frames while improving productivity by reducing the total number of calculations required for visual odometry
Solution Approach 2:
The patent applies partial action by performing dense optical flow calculations only on selected key frames rather than all frames in the video stream. This partial processing maintains sufficient accuracy for vehicle motion determination while significantly reducing computational load and processing time
2Measurement precision
If dense optical flow calculations are used for visual odometry, then measurement precision is improved, but loss of time is worsened due to computational intensity
Solution Approach 1:
By segmenting the video stream and processing only key frames, the patent reduces the time required for visual odometry calculations while maintaining measurement precision through strategic selection of representative frames that capture essential motion information
Solution Approach 2:
The patent implements preliminary action by pre-identifying and selecting key frames from the video stream before performing dense optical flow calculations. This preliminary selection process reduces subsequent calculation time while ensuring that frames with sufficient information content are processed for accurate vehicle motion determination
Data Source
AI summary
A computer, including a processor and a memory, the memory including instructions to be executed by the processor to determine an eccentricity map based on video image data and determine vehicle motion data by processing the eccentricity map and two red, green, blue (RGB) video images with a deep neural network trained to output vehicle motion data in global coordinates. The instructions can further include instructions to operate a vehicle based on the vehicle motion data.