Vehicle Surroundings Detection With Split Neural Network Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle surroundings detection systems consume significant computing power and energy due to the processing of large amounts of data, which is undesirable for on-board vehicle systems.
Innovation Solution
A distributed deep learning architecture is implemented, where a deep neural network is divided into multiple blocks, with some layers processed locally near sensors and others processed remotely in a backend or data center, reducing data transmission and energy consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If all layers of the deep neural network are processed locally on the vehicle, then detection accuracy is improved, but computing power consumption and energy expenditure increase significantly
Solution Approach 1:
The deep neural network is segmented into multiple blocks, with the first block processed locally on the vehicle and the second block processed remotely on a server. This segmentation divides the computational workload, allowing accurate detection while reducing on-vehicle energy consumption by offloading the more computationally intensive second block to a remote server with greater processing power.
2Measurement precision
If large amounts of raw sensor data are transmitted to a remote server for processing, then detection accuracy is improved, but data transmission time and bandwidth consumption increase
Solution Approach 1:
The first block of the deep neural network is executed in advance on the vehicle before data transmission, performing preliminary processing on the sensor data. This preliminary action reduces the amount of data that needs to be transmitted to the remote server, thereby decreasing transmission time and bandwidth consumption while still maintaining high detection accuracy through subsequent processing of the pre-processed data.
3Productivity
If a distributed architecture with multiple computing devices is used, then computing efficiency is improved, but system complexity increases
Solution Approach 1:
The system is segmented into two distinct computing components: a local vehicle unit that executes the first block of the neural network and a remote server that executes the second block. This segmentation improves computing efficiency by distributing the workload to appropriate processing environments while managing complexity through clear functional separation and standardized communication interfaces between the blocks.
Data Source
AI summary
A system for detecting the surroundings of a vehicle includes a sensor module and a plurality of computing devices which are spatially separated from one another and which are configured to process the detected surroundings data by way of an artificial neural network. The artificial neural network includes multiple layers of artificial neurons. Each computing device implements a partial number of the layers.