Autonomous Vehicle Scene Transfer via Semantic Difference Rendering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in efficiently communicating and storing large volumes of image data, which is difficult and costly due to the high volume of acquired image data, necessitating a method to reduce data transfer and storage requirements.
Innovation Solution
The system uses imaging sensors to identify relevant objects in the environment, generating a list of these objects along with their descriptions, which is then transferred over a network, allowing a remote server to synthesize an approximation of the original image by rendering the objects on a predefined map, thereby reducing data volume through subtraction and compression of background objects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If full image data is transferred and stored, then complete environmental information is preserved, but data transfer and storage costs increase significantly
Solution Approach 1:
The patent extracts only the essential semantic information (object identities, positions, and attributes) from complete image data, separating critical environmental information from redundant visual details. This extraction process isolates the necessary data elements for autonomous vehicle operation while discarding unnecessary pixel-level information, thereby resolving the contradiction between information completeness and data volume.
Solution Approach 2:
Instead of transferring and storing actual image data, the system creates simplified semantic representations or models that copy only the essential environmental features. These semantic copies contain object identities, positions, and attributes needed for navigation and decision-making, replacing bulky image files with compact data structures that preserve functional information while dramatically reducing storage requirements.
2Measurement precision
If detailed image data is transferred over the network, then accurate environmental representation is achieved, but network bandwidth consumption increases
Solution Approach 1:
The system extracts only the critical semantic parameters (object types, positions, velocities, and relevant attributes) from complete image data before transmission. This extraction eliminates redundant visual information while preserving the essential environmental representation needed for autonomous vehicle operation, thereby maintaining accuracy while reducing bandwidth consumption.
Solution Approach 2:
The patent transforms image data from pixel-space representation to semantic-parameter representation, changing the data format from detailed visual information to structured parameters describing environmental objects. This parameter transformation maintains the functional accuracy needed for navigation and collision avoidance while dramatically reducing the data size required for network transmission.
3Speed
If complete sensor data is processed locally, then real-time decision making is enabled, but computational resources and power consumption increase
Solution Approach 1:
The patent segments the computational workload between the autonomous vehicle and remote servers. The vehicle performs local processing to extract semantic information and identify environmental objects, then transmits only this processed data to remote servers for further analysis and decision support. This segmentation allows real-time local responses while distributing heavy computational tasks to remote infrastructure, reducing on-vehicle energy consumption.
Solution Approach 2:
The system performs preliminary processing of sensor data on the vehicle to extract essential semantic information before transmission. By pre-processing the data to identify objects, their positions, and attributes, the system reduces the computational burden on both the vehicle and remote servers, enabling faster decision-making with lower energy consumption.
Data Source
AI summary
A system includes at least one imaging sensor and a processor. The processor is configured to acquire detected data describing an environment of an autonomous vehicle using the imaging sensor; derive reference data which describes the environment from a predefined map; compute difference data representing a difference between the detected data and the reference data; and transfer the difference data, wherein an image computed based on the difference data and the reference data represents the detected data. Other embodiments are also described.