Control Output Probability Estimation With Certainty Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems struggle to accurately calculate and quantify the probability distribution and reliability of control outputs in complex and safety-critical robotic systems with limited data, often leading to overconfidence in probability estimation.
Innovation Solution
A computation system that calculates probability distributions by discretely labeling outputs, estimating pseudo-data counts, and using non-parametric Bayes to model and learn pseudo-count numbers, correcting probabilities based on certainty, thereby suppressing overconfidence and improving reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If probability distribution is calculated using limited data in complex robotic systems, then calculation speed is improved, but measurement precision of probability distribution deteriorates
Solution Approach 1:
The patent replaces traditional frequentist probability methods with Bayesian probability theory to calculate probability distributions from limited data. The Bayesian approach allows incorporation of prior knowledge and uncertainty quantification, enabling reliable probability distribution estimation even with small sample sizes in complex robotic systems.
Solution Approach 2:
The patent transforms the probability calculation by introducing certainty of probability as an additional parameter. This parameter adjusts the probability distribution based on the reliability of the input data, allowing the system to maintain calculation speed while improving measurement precision through uncertainty-aware probability estimation.
2Device complexity
If probability distribution is calculated without considering uncertainty, then calculation complexity is reduced, but reliability of probability estimation deteriorates
Solution Approach 1:
The patent introduces a feedback mechanism where the certainty of probability is calculated based on the input data and used to adjust the final probability distribution. This feedback loop ensures that the reliability of probability estimation is maintained by down-weighting uncertain predictions, without requiring complex recalculations of the entire probability model.
Solution Approach 2:
The patent introduces certainty of probability as an intermediary parameter that mediates between the raw probability distribution and the final reliable probability estimation. This intermediary allows the system to maintain simple calculation structures while improving reliability through uncertainty-aware adjustment of probability values.
3Productivity
If probability distribution is calculated with high confidence, then decision-making speed is improved, but manufacturing precision of probability estimation deteriorates
Solution Approach 1:
The patent applies partial correction to the probability distribution by adjusting only the uncertain portions using certainty of probability. This allows the system to maintain high decision-making speed for confident predictions while applying precision corrections only where needed, avoiding excessive computation across the entire probability distribution.
Data Source
AI summary
In a computation system for calculating a probability distribution of an output of control for each input condition in control of an object, the computation system according to the present disclosure is configured to perform calculation of the probability distribution of the output of the control performed by calculating certainty of probability for each probability distribution interval and correcting the probability for each interval according to a degree of certainty of the probability. In this correction, the probability distribution of the output of control is calculated by correcting the probability for each interval so that a probability reduction rate when the certainty of probability is less than a predetermined value is larger than a probability reduction rate when the certainty of probability is higher than a predetermined value.