Preceding Vehicle Scene Prediction for Fuel-Saving Driving Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional driving assistance technologies face reduced prediction accuracy and fuel efficiency due to mismatched learning and input data sets, and inadequate consideration of prediction model ranges, leading to incorrect fuel-saving driving control.
Innovation Solution
A driving assistance device that detects the distance to a preceding vehicle's predetermined point, determines the traveling scene, and selects a prediction model based on that scene to calculate the vehicle's state after a predetermined time, allowing for improved fuel efficiency by controlling the own vehicle's speed accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single prediction model is used for all traveling situations, then the device complexity is reduced, but the prediction accuracy deteriorates when the current situation differs from the learning data set
Solution Approach 1:
The patent segments the prediction model into multiple scene-specific models (intersection scene model, non-intersection scene model) based on the traveling situation. The determination unit classifies the current scene type, and the calculation unit selects the appropriate prediction model accordingly. This segmentation allows each model to specialize in its specific scene, improving prediction accuracy without significantly increasing overall system complexity.
2Use of energy by moving object
If fuel-saving driving control is executed based on prediction in all situations, then fuel efficiency is improved, but the reliability deteriorates when the prediction model is applied outside its assumed range
Solution Approach 1:
The patent introduces dynamic selection of prediction models based on the determined traveling scene. The calculation unit dynamically switches between different prediction models (intersection-specific or non-intersection-specific) according to the current situation determined by the determination unit. This dynamic adaptation ensures that the prediction is always based on the most appropriate model for the current context, maintaining both fuel efficiency and reliability.
3Ease of operation
If the prediction model range is not considered, then the ease of operation is improved, but the prediction accuracy deteriorates due to mismatched learning and input data sets
Solution Approach 1:
The patent performs preliminary classification of the traveling scene type before executing the prediction. The determination unit first determines whether the current situation is an intersection scene or non-intersection scene, and only then does the calculation unit select and apply the appropriate prediction model. This preliminary action ensures that the prediction is applied within the correct model range, maintaining accuracy while keeping the overall process automated and simple to operate.
Data Source
AI summary
A driving assistance device controls the speed of an own vehicle based on a predicted traveling state of a preceding vehicle, and includes a traveling scene recognition unit that detects the distance until the preceding vehicle reaches an intersection, a determination unit that determines a traveling scene of the preceding vehicle at the intersection, and a calculation unit that selects the preceding vehicle prediction model according to the traveling scene determined by the determination unit from among a plurality of preceding vehicle prediction models recorded for each traveling scene in which the preceding vehicle travels, and inputs a distance until the preceding vehicle detected by the traveling scene recognition unit in the selected preceding vehicle prediction model reaches the intersection so as to calculate a predicted traveling state of the preceding vehicle after a predetermined time.