Autonomous Driving Model Debugging via Sub-sampled Frame Transmission
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional autonomous driving systems face delays in model updates and error detection due to limited wireless bandwidth, which hinders real-time analysis and rapid deployment of improvements, as they require high-bandwidth connections for uploading high-resolution video and classification data.
Innovation Solution
The method involves sub-sampling video frames and transmitting sub-sampled frames along with classification data to a remote device for real-time analysis, allowing for immediate adjustments and updates to the autonomous driving system, enabling rapid model updates and error correction while the vehicle is operational.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If high-resolution video and classification data are transmitted via wireless network while the autonomous agent is active, then real-time analysis and rapid model updates are enabled, but wireless network bandwidth requirements become prohibitively high
Solution Approach 1:
The patent extracts and transmits only the essential elements needed for model improvement - specifically sub-sampled video frames and classification data - rather than transmitting complete high-resolution video. This extraction approach enables real-time wireless transmission while significantly reducing bandwidth requirements.
Solution Approach 2:
The patent changes the resolution parameter of video frames by applying sub-sampling techniques. This parameter modification reduces the data size of video frames while preserving sufficient information for model analysis, thereby enabling real-time transmission over wireless networks with limited bandwidth.
2Measurement precision
If complete high-resolution video frames are transmitted, then analysis accuracy is maintained, but transmission time and bandwidth consumption increase
Solution Approach 1:
The patent creates a simplified copy of the video data through sub-sampling, producing reduced-resolution frames that retain the essential information needed for classification accuracy. This copy approach maintains measurement precision while dramatically reducing transmission time.
3Reliability
If video and classification data are uploaded only when the autonomous agent is at a location with high bandwidth connection, then data transmission reliability is improved, but the time delay between data generation and model improvement increases
Solution Approach 1:
The patent transforms the static upload approach into a dynamic transmission system. Instead of waiting for high-bandwidth connections, the system continuously transmits sub-sampled data over wireless networks with varying bandwidth conditions, adapting to available network capacity in real-time.
Solution Approach 2:
The patent performs preliminary sub-sampling and preparation of video data before transmission. By pre-processing the video frames to reduced resolution, the system prepares data that can be transmitted reliably over wireless networks regardless of instantaneous bandwidth availability, reducing the need to wait for optimal connection conditions.
Data Source
AI summary
A method for improving an autonomous driving system for an autonomous vehicle is disclosed. The method includes sub-sampling a frame generated by an output of a sensor and transmitting, to a remote device, the sub-sampled frame and classification data corresponding to the sub-sampled frame. The method also includes receiving, from the remote device, an adjustment to the autonomous driving system in response to the transmitted sub-sampled frame and classification data. The method further includes controlling an action of the autonomous vehicle based on the adjusted autonomous driving system.