Interaction-Stiffness-Guided Basis Allocation in Dynamic Movement Primitives for Efficient Skill Transfer

2026-10-01

AI Summary

This study addresses the limitation of predefined basis function layouts in Dynamic Movement Primitives (DMPs), which struggle to accommodate varying precision requirements across different trajectory phases. To overcome this, we propose SC-DMPs, a method that constructs a criticality index by integrating operator-robot interaction stiffness with trajectory consistency. Through inverse cumulative criticality mapping, SC-DMPs dynamically reallocates the centers and bandwidths of basis functions within the canonical phase domain. Experimental evaluations on handwriting and real-world robotic tasks demonstrate that SC-DMPs significantly enhances trajectory reproduction, endpoint generalization, and precision during critical phases while preserving model compactness and stability. This work establishes a novel paradigm for adaptive, high-precision skill learning.

Abstract

Dynamic Movement Primitives (DMPs) provide a compact and stable formulation for trajectory representation and generalization in robot skill learning. However, their predefined basis layout limits the allocation of approximation capacity according to stage-dependent precision requirements. To address this issue, this article proposes Stage-Criticality-Guided Dynamic Movement Primitives (SC-DMPs) with adaptive basis allocation for precision-critical skill learning. Operator-robot interaction stiffness and a trajectory-consistency cue derived from cross-demonstration task-space variability are integrated to construct a stage-criticality index. Guided by this index, basis centers are redistributed in normalized time through inverse cumulative criticality and mapped to the canonical phase domain, while their bandwidths are refined to adjust local approximation support. This enables denser and more flexible representation at high-criticality stages while retaining sparser allocation elsewhere. Experiments on handwriting trajectories and three real-robot tasks show that the inferred criticality is concentrated in geometrically demanding and task-constrained regions. Comparisons with DMPs, ProMPs, ProDMP, GP-MP, and KMP demonstrate improved trajectory reproduction, endpoint generalization, and task-critical accuracy while retaining a compact model and the stable structure of classical DMPs.

Problem

• Dynamic Movement Primitives
• Basis Allocation
• Skill Transfer
• Trajectory Representation
• Interaction Stiffness

Innovation

• Dynamic Movement Primitives
• Adaptive Basis Allocation
• Interaction Stiffness
• Stage-Criticality Index
• Skill Transfer