Vehicle Speed Profile Control Using ML and Dynamic Programming
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle speed optimization systems fail to efficiently consider various driving factors, including real-time traffic conditions and road characteristics, leading to suboptimal energy consumption and stability in vehicles.
Innovation Solution
An intelligent vehicle driving control system that utilizes a combination of machine learning models, such as CNN and RNN, and dynamic programming to derive an optimal speed profile by analyzing wheel torque, speed, and traffic information, minimizing energy consumption and improving fuel efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If intensive data processing on driving road shapes and traffic conditions is performed in a large capacity cloud system, then global optimal speed profile can be calculated, but high computational power and system complexity are required
Solution Approach 1:
The system segments the optimal speed profile calculation into two parts: (1) cloud-based processing that collects and processes driving data, road shapes, and traffic conditions to generate optimization models, and (2) vehicle-based execution that applies pre-calculated optimization algorithms and speed profiles. This segmentation reduces cloud system complexity while maintaining calculation accuracy.
Solution Approach 2:
The cloud system performs preliminary actions by pre-calculating optimization models and speed profiles based on historical data, road shapes, and traffic patterns before vehicles need them. These pre-computed models are then transmitted to vehicles, reducing real-time computational requirements and system complexity during actual driving operations.
2Loss of energy
If machine learning models and dynamic programming methods are used to calculate optimal speed profiles, then energy consumption is minimized, but computational processing time and complexity increase
Solution Approach 1:
Optimization algorithms including machine learning models and dynamic programming methods are executed in advance by the cloud system to generate optimal speed profiles. These pre-computed profiles are then transmitted to vehicles for direct execution, minimizing real-time computational processing time while maintaining energy optimization benefits.
Solution Approach 2:
The system replaces complex real-time mechanical computation in vehicles with pre-computed digital optimization models transmitted from the cloud. This substitution allows sophisticated energy optimization algorithms to run in the cloud environment rather than requiring equivalent computational power in each vehicle's real-time control system.
Data Source
AI summary
In one aspect, an operating method of an intelligence vehicle driving control system is provided that comprises: a collecting step of collecting big data including a wheel torque and a speed for every vehicle type and traffic information; a torque calculating step of learning the big data using a predetermined machine learning model and inputs a specific desired speed profile to the machine learning model to calculate a motor torque of a driving vehicle; and an optimal speed profile deriving step of calculating an energy consumption required to generate the calculated motor torque using a predetermined dynamic programming method and a reverse vehicle dynamic model and deriving an optimal speed profile in which the energy consumption is minimized.