Dynamic Movement Primitive State Estimation for High-Dimensional Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing state estimation methods, such as Kalman filters, struggle to scale accurately to high-dimensional problems with many nonlinear interactions, which are common in fields like robotic control and gesture recognition.
Innovation Solution
The use of Dynamic Movement Primitives (DMPs) to decompose the state estimation problem into simpler components, where the state is generated by a DMP, allowing for the estimation of a low-dimensional state and classification of signals, implemented using a computer system with a point attractor and a forcing function, and estimated by an artificial neural network.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Kalman filters are used for state estimation, then linear dynamical systems can be optimally estimated, but they struggle with high-dimensional nonlinear problems
Solution Approach 1:
The patent segments the high-dimensional state estimation problem into multiple low-dimensional subproblems by decomposing the state space into clusters. Each cluster is handled by a separate Kalman filter instance, allowing the system to maintain optimality for linear dynamics while scaling to high-dimensional nonlinear problems through distributed processing.
Solution Approach 2:
The patent transitions from directly estimating the full high-dimensional state to estimating a lower-dimensional representation (cluster assignments and cluster centers). This dimensional reduction allows Kalman filters to operate effectively while still capturing the essential dynamics of the high-dimensional system through the clustered structure.
2Adaptability or versatility
If the state space is increased to handle more complex gestures, then gesture recognition capability improves, but computational complexity increases
Solution Approach 1:
The patent segments the large state space into multiple smaller clusters, allowing the system to handle complex gestures by distributing the computational load across multiple simpler sub-problems. Each cluster represents a localized region of the state space that can be processed more efficiently.
Solution Approach 2:
The patent uses a fixed number of clusters that may be sufficient for most gestures rather than exhaustively covering every possible state. This partial coverage approach reduces computational complexity while maintaining adequate gesture recognition capability for practical applications.
Data Source
AI summary
A system continuously estimating the state of a dynamical system and classifying signals comprising a computer processor and a computer readable medium having computer executable instructions for providing: a module estimating of the state of a dynamical system assumed to be generated by a Dynamic Movement Primitive; a module classifying signals through inspecting dynamical system state estimates; and a coupling between the two modules such that classifications reset the dynamical system state estimate.