The invention belongs to the technical field of
robot structure design and intelligent optimization, and particularly discloses a
micro motor direct-driven dexterous hand double-layer optimization design method, which realizes configuration and proportion parameter collaborative optimization through two-stage optimization: firstly, extracting the geometric dimension of a dexterous hand as a
gene variable, setting a physical boundary constraint, and then, carrying out two-stage optimization; constructing a parameterized model
library in combination with normalized mapping and template rendering technologies; through Monte Carlo random sampling, convex
hull volume calculation and a CMA-ES
algorithm, an optimal macroscopic configuration is screened from multiple
thumb distribution and motor integration
modes; and finally, introducing a human action data sequence, and finely adjusting microscale parameters through a double-layer architecture of outer-layer parameter search and inner-layer
pose matching. According to the method, the hardware
trial and error cost is greatly reduced, the cooperation
work space of the dexterous hand is expanded, the personification
operation safety and the physical feasibility are improved, and the method is suitable for direct-drive dexterous hand research and development in the fields of industry, service and
rehabilitation robots.