Method for controlling a robot and robot controller

By reducing the dimensionality of the high-dimensional non-Euclidean parameter space and introducing geometry-aware Bayesian optimization, the problem of performance degradation of Bayesian optimization in high-dimensional parameter space is solved, and faster convergence, higher accuracy and lower computational cost are achieved.

CN112987563BActive Publication Date: 2025-09-23ROBERT BOSCH GMBH
0 Cites 0 Cited by

Patent Information

Application Number
CN202011472999.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Priority Date
2019-12-16
Filing Date
2020-12-15
Publication Date
2025-09-23
Estimated Expiration
2040-12-15

AI Technical Summary

Technical Problem

The performance of Bayesian optimization for optimizing robot control parameters in high-dimensional non-Euclidean parameter spaces is severely affected, resulting in degraded performance and increased computational cost.

Method used

By transforming the original control parameter space into a reduced control parameter space, the high-dimensional parameter space is reduced in dimensionality using a geometry-aware Bayesian optimization method, including techniques such as the principal nested sphere algorithm, the orthogonal projection algorithm, and the Gaussian mixture model, and the acquisition function is optimized in the reduced control parameter space using the conjugate gradient method and the geometry-aware kernel for optimization.

Benefits of technology

Improved performance and scalability of Bayesian optimization, achieving faster convergence, better accuracy, lower solution variance, and reduced computational cost.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

According to various embodiments, a method for controlling a robot using control parameter values ​​from a non-Euclidean original control parameter space is described, comprising: performing a Bayesian optimization of an objective function representing a desired control objective for the robot on the original control parameter space; and controlling the robot according to the control parameter values ​​from the original control parameter space found in the Bayesian optimization; wherein the Bayesian optimization comprises: transforming the original control parameter space into a reduced control parameter space using the observed control parameter values, wherein the original control parameter space comprises a first number of dimensions, wherein the reduced control parameter space comprises a second number of dimensions, and wherein the first number of dimensions is higher than the second number of dimensions; determining an evaluation point of the objective function in the reduced control parameter space by searching for an optimal value of an acquisition function in an iterative search, comprising, at each iteration, updating a candidate evaluation point using a search direction in a tangent space of the reduced control parameter space at the candidate evaluation point; mapping the updated candidate evaluation point from the tangent space to the reduced control parameter space; and using the mapped updated candidate evaluation point as an evaluation point for a next iteration until a stopping criterion is met; and mapping the determined evaluation point from the reduced control parameter space to the original control parameter space.
Need to check novelty before this filing date? Find Prior Art