Robotic Motion Retargeting for Real-Time Vibration Suppression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing robotic systems face challenges in suppressing unwanted vibrations during rapid movements, leading to undesirable oscillations that compromise the accuracy and realism of animations, particularly in audio-animatronic figures and other robotic characters, due to the inherent compliance and deformation in their components and joints.
Innovation Solution
A real-time computational vibration suppression method using state-of-the-art machine learning techniques in conjunction with a differentiable dynamics simulator to generate control signals that minimize vibrations, allowing for the use of lighter and less expensive robotic systems by optimizing motor trajectories and accounting for the flexibility of components and actuators.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If robotic systems are designed to be as stiff as possible to reduce vibrations, then vibration suppression is improved, but the system becomes heavier and more expensive
Solution Approach 1:
The patent replaces mechanical stiffness (physical property) with computational vibration suppression (software-based control). The system uses a machine learning model trained on dynamic simulation data to predict and compensate for vibrations through control signals, eliminating the need for heavy stiffening structures while achieving vibration reduction.
Solution Approach 2:
The patent changes the control parameters in real-time based on predicted vibration characteristics. The machine learning model outputs adjusted control parameters that compensate for anticipated vibrations, allowing the system to maintain flexibility while suppressing unwanted oscillations through parameter optimization rather than structural rigidity.
2Object-affected harmful factors
If robotic systems are designed to be as stiff as possible to reduce vibrations, then vibration suppression is improved, but the system becomes more expensive
Solution Approach 1:
The patent replaces expensive mechanical stiffening solutions with a computational control system. By using machine learning-based vibration prediction and compensation, the system achieves vibration suppression without requiring costly rigid structures, specialized materials, or precision-machined components.
Solution Approach 2:
The patent uses lighter, less expensive components that would normally be prone to vibration, compensating for their deficiencies through software-based control. Instead of investing in expensive rigid components, the system uses affordable lightweight materials and relies on the machine learning controller to maintain performance.
3Object-affected harmful factors
If manual tuning of motion trajectories is performed to avoid vibrations, then vibration suppression is improved, but the design process becomes more time-consuming
Solution Approach 1:
The patent implements self-service vibration suppression through an autonomous machine learning model. The system automatically predicts vibrations and generates compensatory control signals without requiring manual intervention. The model learns from simulation data and independently optimizes control parameters, eliminating the need for designers to manually tune trajectories to avoid vibrations.
Solution Approach 2:
The patent performs preliminary vibration prediction using the trained machine learning model before executing movements. The system anticipates upcoming vibrations based on the planned motion trajectory and pre-computes compensation signals, allowing vibration suppression without real-time trial and error or manual tuning during the design process.
4Object-affected harmful factors
If computational vibration techniques are performed offline in the design stage, then vibration suppression is improved, but real-time control capability is reduced
Solution Approach 1:
The patent transitions from static offline computation to dynamic real-time control. The machine learning model is trained offline on simulation data but then deployed for real-time inference during robot operation. The system dynamically adjusts control parameters based on current and predicted states, enabling adaptive vibration suppression that responds to changing conditions while maintaining fast control loop speeds.
Solution Approach 2:
The patent performs preliminary training of the machine learning model offline using dynamic simulation data, then deploys the trained model for rapid real-time prediction and control. The heavy computational work of learning vibration patterns is done in advance, allowing the system to make fast predictions during operation without requiring real-time simulation computation.
Data Source
AI summary
In one example, a robotic system is disclosed that includes a plurality of components coupled together, a plurality of motors operable to move the plurality of components, a controller in electrical communication with the plurality of motors to generate control signals to actuate movement of the plurality of components, wherein the controller is configured to: receive a first set of control signals operative to generate a defined motion for the plurality of components, analyze the first set of control signals to determine a second set of control signals operative to define a retargeted motion for the plurality of components, wherein the retargeted motion suppresses vibrations of the plurality of components as compared to the defined motion, and provide the second set of control signals to the plurality of motors to actuate the retargeted motion by the plurality of components.