Handlebar Stabilizing Torque Control for Low-Speed Vehicle Balance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle stabilizing systems face challenges in maintaining stability, especially at low speeds, due to the use of complex and costly sensor configurations that provide inaccurate and delayed steering assistance, which can lead to rider fatigue and increased risk of capsize failure.
Innovation Solution
A vehicle stabilizing system that utilizes a simplified sensor setup comprising a position sensor, angular displacement sensor, and steering torque sensor to determine a synthesized torque based on roll angle and roll rate, applying a stabilizing torque through an actuator to mimic rider input, with consideration for vehicle speed and delay times.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If electronic stability control systems use traditional sensor fusion algorithms, then processing speed is limited, but system complexity and computational load increase
Solution Approach 1:
The patent replaces traditional mechanical sensor fusion algorithms with a neural network-based system. The neural network is trained offline to learn optimal fusion strategies, then deployed for real-time processing. This substitution of computational mechanics with pre-trained intelligent models achieves faster processing speeds while reducing real-time computational complexity and resource requirements.
2Measurement precision
If multiple sensors are integrated to improve detection accuracy, then measurement precision improves, but device complexity increases
Solution Approach 1:
The neural network system performs self-optimization by automatically learning the optimal weighting and fusion strategies for multiple sensors during the training phase. The system adapts to different sensor conditions and configurations without requiring manual calibration or complex real-time adjustment mechanisms, thereby maintaining high detection accuracy while minimizing system complexity.
Solution Approach 2:
The patent performs preliminary training of the neural network offline before deployment. During this preliminary action, the system learns optimal sensor fusion strategies from extensive training data, including various driving conditions and sensor failure scenarios. This pre-computed knowledge is then applied during real-time operation, achieving high measurement precision without the need for complex real-time computational processes.
3Productivity
If traditional sensor fusion algorithms are used, then computational resources are heavily consumed, but processing speed is limited
Solution Approach 1:
The patent replaces computationally intensive traditional sensor fusion algorithms with a neural network model that has been pre-trained offline. The neural network performs inference operations that are significantly less computationally demanding than traditional iterative optimization algorithms, thereby increasing processing throughput while reducing real-time energy consumption and computational resource requirements.
Data Source
Figure 1
Figure 2
Figure 3~4a
AI summary
Aspects of stabilizing a vehicle are described. A vehicle speed (v), a roll angle (ϕ), and a roll rate (͘ϕ) of the vehicle are determined. Gain values (G1, G2) are determined respectively for the roll angle (ϕ) and the roll rate (͘ϕ) based on the vehicle speed (v); and a stabilizing torque (Ts) is determined based on application of respective gain values (G1, G2) to the roll angle (ϕ) and the roll rate (͘ϕ). In one example, a tuning parameter may be determined based on a comparison of a synthesized torque (T) with a steering torque (Tr) and may be additionally used to determine the stabilizing torque (Ts). An actuating signal corresponding to a stabilizing torque (Ts) to be applied to a steering handle of the vehicle by the actuator is provided to an actuator for stabilizing the vehicle.