EV Travelable Area Prediction Using Terrain-Aware Power Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies fail to accurately generate travelable area information for electrically driven mobile vehicles due to inflexible calculation methods that do not account for changes in vehicle performance over time.
Innovation Solution
An information processing method using a learned model obtained through machine learning to calculate geographical feature quantities and electric power consumption between locations, incorporating traveling history data to update the model and accurately determine travelable areas based on current vehicle performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If calculation of travelable area is performed on a rule basis, then calculation process is simple, but generation of travelable area information accurately indicating the travelable area fails
Solution Approach 1:
The patent applies dynamics by transitioning from static rule-based calculation to dynamic machine learning model that adapts to changing vehicle performance characteristics. The learned model updates based on actual travel data, enabling accurate travelable area calculation that reflects current vehicle state rather than relying on fixed rules.
Solution Approach 2:
The patent changes the calculation parameters from simple rule-based metrics to complex geographical feature quantities including elevation, slope, curvature, and road grade. These detailed parameters are input into the machine learning model to achieve accurate travelable area prediction that accounts for varying terrain and vehicle performance conditions.
2Measurement precision
If machine learning model is used to calculate electric power consumption, then travelable area information accuracy is improved, but processing complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-training the machine learning model offline using historical travel data before deployment. The model is prepared in advance with learned relationships between geographical features and power consumption, so that during actual operation, travelable area calculation uses the pre-trained model efficiently without requiring real-time complex processing.
3Productivity
If rule-based calculation method is used, then processing cost is low, but travelable area information does not reflect latest vehicle performance
Solution Approach 1:
The patent implements feedback by using actual travel data and power consumption measurements to continuously update and retrain the machine learning model. This feedback loop ensures the model learns from real vehicle performance data and adapts to changes in vehicle characteristics over time, maintaining accurate travelable area calculations that reflect the latest performance state.
Data Source
AI summary
An information processing device performs: acquiring input data including a current location and a residual electric power amount of an electrically driven mobile vehicle; calculating, on the basis of map information, a geographical feature quantity between the current location and a plurality of locations; calculating an estimative electric power consumptive amount of the electrically driven mobile vehicle from the current location to each of the locations by inputting the geographical feature quantity into a learned model obtained by machine learning of a relationship between: a geographical feature quantity between two locations; and an electric power consumption amount having been consumed to travel between the two locations; generating, on the basis of the estimative electric power consumptive amount for each of the locations and the residual electric power amount, travelable area information indicating a travelable area for the electrically driven mobile vehicle from the current location.