Vehicle Speed Profile Generation Using Statistical Lookup Tables
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for generating a driver's speed profile lack a defined procedure for calibrating models based on actual driving data, leading to unrealistic outcomes, and machine learning-based approaches require substantial computational resources and high costs.
Innovation Solution
A computer-implemented method using statistical analysis to generate a speed profile by collecting data on environmental conditions and vehicle attributes, adjusting speed values based on these attributes, and selecting values based on probability, optionally considering driver profiles, to create a more accurate representation of real-world driving behaviors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine learning techniques are used to generate speed profiles, then accuracy of representing real-world driving behaviors is improved, but computational resources and costs increase substantially
Solution Approach 1:
The patent replaces expensive machine learning models with a simpler, more economical statistical approach using probability distributions and lookup tables. Instead of training complex neural networks that consume substantial computational resources, the system uses pre-computed statistical parameters (mean, standard deviation) derived from actual driving data, which can be stored in compact lookup tables. This achieves comparable accuracy in representing real-world driving behaviors while dramatically reducing computational energy requirements and costs.
2Device complexity
If physical vehicle models are used without driver behavior calibration, then computational simplicity is maintained, but realism and accuracy of speed profiles deteriorate
Solution Approach 1:
The patent performs preliminary calibration by collecting actual driving data and computing statistical parameters (mean and standard deviation of speed variations) before the vehicle operates. These pre-computed statistical characteristics are stored in lookup tables that can be quickly referenced during vehicle operation. This preliminary action embeds real-world driving behavior patterns into the system, ensuring realistic speed profiles while maintaining computational simplicity during actual vehicle operation, as the complex data processing is done beforehand.
3Productivity
If statistical analysis with probability-based selection is used, then computational efficiency is improved, but model flexibility may be reduced
Solution Approach 1:
The patent implements a dynamic statistical model where probability distributions and their parameters (mean, standard deviation) are not fixed but adapt to different driving conditions, vehicle types, and environmental contexts. The system maintains multiple statistical profiles that can be selected or blended based on current operating conditions, allowing the model to flexibly represent diverse driving behaviors while still using computationally efficient statistical methods and lookup tables for rapid evaluation.
Data Source
Figure 1~2
AI summary
A computer-implemented method (20) for automatically generating a speed profile of a vehicle (1) to be driven on a predetermined path, the method comprising the following steps: - collecting a first set of data representative of environmental conditions at waypoints along the predetermined path (21), each distinct environmental condition being associated with alternative speed fluctuation values and each alternative speed fluctuation value being associated with a probability value; - collecting a second set of data representative of at least one vehicle attribute (22); - adjusting said alternative speed fluctuation values based on the at least one vehicle attribute (23); - selecting for each waypoint a speed fluctuation value among the associated alternative speed fluctuation values according to the associated probability values (24); and - generating the speed profile of the vehicle according to the selected speed fluctuation values and their waypoints (25).