Self-Driving Car Path Graph Caching for Real-Time Decisions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The generation of graph-structures for Self-Driving Cars (SDCs) is computationally expensive due to the need to compute multiple attributes for edges, which can lead to delays in decision-making and object detection, particularly in scenarios where fast corrective actions are required, such as avoiding obstacles.
Innovation Solution
The implementation of a parallel processing pipeline that separates the computation of static and dynamic attributes, where static attributes are generated and cached separately from dynamic attributes, allowing for quicker generation of graph-structures and improved computational efficiency during SDC operation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If graph-structure generation computes multiple attributes for edges sequentially, then computation accuracy is improved, but computational time increases
Solution Approach 1:
The patent segments the graph-structure generation process into two independent parallel pipelines: a first processing pipeline that generates static attributes for potential positions, and a second processing pipeline that generates dynamic attributes for edges. This segmentation allows simultaneous computation of different attribute types without interfering with each other, thereby reducing total computational time while maintaining accuracy of individual attribute computations.
Solution Approach 2:
The first processing pipeline performs preliminary computation of static attributes for all potential positions and caches them before the second pipeline needs them. This preliminary action eliminates the need for the second pipeline to wait for static attribute computation, allowing dynamic attribute generation to proceed independently and significantly reducing overall computational time.
2Speed
If graph-structure generation is performed in real-time, then decision-making speed is improved, but computational complexity increases
Solution Approach 1:
The system divides the complex real-time graph-structure generation task into two simpler, independent processing pipelines with distinct functions. The first pipeline handles static attributes (less time-sensitive), while the second handles dynamic attributes (more time-sensitive). This segmentation reduces the computational complexity of each individual pipeline, making real-time execution feasible.
Solution Approach 2:
By pre-computing and caching static attributes before they are needed by the second pipeline, the system eliminates a significant portion of computational complexity from the time-critical path. The second pipeline only needs to compute dynamic attributes using already-available static data, substantially reducing its computational burden and enabling real-time operation.
3Productivity
If static attributes are computed separately and cached, then computational efficiency is improved, but memory usage increases
Solution Approach 1:
The first processing pipeline pre-computes static attributes for all potential positions and stores them in a cache memory structure. This preliminary computation allows the second pipeline to retrieve static attributes instantly without re-computation, dramatically improving computational efficiency. The memory trade-off is acceptable because static attributes are computed once and reused multiple times across different edges and time steps.
Solution Approach 2:
The system creates a cached copy of static attributes that can be rapidly accessed by the second processing pipeline. Instead of repeatedly computing the same static attributes for multiple edges, the system stores them once and references them multiple times, improving computational efficiency while using memory for the cached data.
Data Source
AI summary
A method and device for operating a Self-Driving Car (SDC) are disclosed. The device executes in real-time a first processing pipeline including generating static attributes for a plurality of potential positions of the SDC on the road segment, and caching the static attributes in association with the respective ones of the plurality of potential positions. The device executes in real-time a second processing pipeline in parallel with the first processing pipeline including generating a graph-structure for operating the SDC on the road segment. The generating the graph-structure includes generating dynamic attributes for a given edge of the graph-structure, acquiring from the cache memory static attributes for the given edge of the graph-structure, such that the given edge in the graph-structure is associated with the static attributes generated by the first processing pipeline and with the dynamic attributes generated by the second processing pipeline.