AI Prosthesis Tuning for User-Preferred Ankle Stiffness
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing robotic assistive prostheses require a time-consuming tuning process to match stiffness to individual user preferences, which is challenging to scale and may not account for shifting user objectives during different activities.
Innovation Solution
A machine learning approach using biomechanical data to predict user-preferred ankle stiffness, employing algorithms like LSTM trained on user-specific data to automate the tuning process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a manual tuning process is used to match prosthesis stiffness to user preference, then the prosthesis can be customized to individual user needs, but the process becomes time-consuming and difficult to scale
Solution Approach 1:
The system performs preliminary actions by collecting biomechanical data during normal prosthesis use and pre-processing this data to establish user-specific patterns. The machine learning model is trained in advance on this data, so that when tuning is needed, the system can quickly predict optimal stiffness settings without requiring lengthy manual adjustment sessions. This preliminary data collection and model training resolves the contradiction by preparing the customization information beforehand.
Solution Approach 2:
The prosthesis system performs self-service by automatically collecting its own biomechanical data during operation, using this data to train machine learning models that predict user preferences, and then autonomously determining optimal stiffness settings. This self-service capability eliminates the need for time-consuming manual tuning while maintaining high adaptability to individual user needs, as the system continuously learns from and adapts to each user's unique biomechanical patterns.
2Measurement precision
If extensive user-specific training data is collected to improve prediction accuracy, then the model can better capture individual user preferences, but the data collection process becomes more complex and time-consuming
Solution Approach 1:
The system uses multi-functionality by employing a single biomechanical sensor suite on the prosthesis to serve multiple purposes: it collects data for real-time control, captures user preference information, and provides training data for machine learning models. This universal data collection approach improves prediction accuracy without requiring separate specialized sensors or additional complex data collection apparatus, thereby resolving the contradiction between precision and complexity.
Solution Approach 2:
The system implements continuous data collection during normal prosthesis operation, transforming routine usage into an ongoing training process. Rather than requiring separate dedicated training sessions that would increase complexity and time requirements, the system continuously accumulates user-specific biomechanical data during everyday use. This continuous action approach improves prediction accuracy over time while maintaining simple, unobtrusive data collection that doesn't disrupt normal operation.
Data Source
AI summary
Techniques are provided for using machine learning methods to predict user preferences with respect to robotic assistive prostheses and/or custom tune the robotic assistive prostheses for individual users based on the predicted user preferences. A training biomechanical dataset, including historical biomechanical sensor data for a robotic assistive prosthetic device operating using a plurality of tuning settings at a plurality of speeds, and a training user preference dataset including historical user tuning preference data for the robotic assistive prosthetic device for each respective tuning setting and speed, are generated. A machine learning model is trained using the training biomechanical dataset and the training user preference dataset. The trained machine learning model is applied to new biomechanical sensor data associated with a particular user to predict the user's tuning preferences for the robotic assistive prosthetic device, and the settings of the device are automatically modified based on the predicted tuning preferences.