Automotive HUD Image Warping Using an NPU Instead of a GPU
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional automotive HUD systems rely on graphical processing units (GPUs) for image warping, which increases processing load and costs as more displays are added, necessitating larger or multiple GPUs.
Innovation Solution
Utilize a neural processing unit (NPU) to perform machine learning-based image warping, generating and training a warping model that can be executed by the NPU, freeing up GPU resources for other graphical processing tasks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a graphical processing unit (GPU) is used to warp the image for the HUD system, then the image warping function is achieved, but the processing load on the GPU increases and hardware costs increase
Solution Approach 1:
The image warping function is extracted from the GPU and implemented as a standalone neural network model that can be executed by general-purpose processors. This separates the warping task from the graphical processing workload, reducing GPU complexity and cost while maintaining the required image warping capability for the HUD system
Solution Approach 2:
The neural network warping model can be deployed on various processing platforms (CPU, GPU, NPU, or other general-purpose processors), making the warping function universal and not dependent on dedicated GPU hardware. This multi-platform compatibility reduces hardware complexity and allows flexible deployment
2Productivity
If the processing load on the GPU increases due to adding more displays, then more graphical processing capability is available, but larger or multiple GPUs are required at increased costs
Solution Approach 1:
By extracting the image warping task from the GPU and implementing it as an independent neural network model executable on general-purpose processors, the GPU is relieved of non-graphical processing tasks. This allows the GPU to focus solely on graphical rendering, improving efficiency and reducing the need for additional or larger GPU hardware
Solution Approach 2:
A neural network model serves as an intermediary between the image processing requirements and the GPU. The model preprocesses images using general-purpose processors, then passes the processed results to the GPU for final rendering, enabling efficient distribution of processing tasks and reducing overall system complexity
Data Source
AI summary
A heads-up display (HUD) system for an automobile includes a projection system configured to project a warped image onto a reflective portion of a surface of a windshield of the automobile, wherein the windshield surface defines a curvature such that the reflected projected warped image appears substantially non-warped to a driver of the automobile and a control system comprising a neural processing unit (NPU) configured to execute a set of machine learning based tasks of the automobile, including obtaining a trained warping model configured for warping an image to obtain the warped image and executing the trained warping model on the image to obtain the warped image, wherein the control system does not utilize a graphical processing unit (GPU) to warp the image or to otherwise obtain the warped image.