Human-Guided Image Feature-Based Visual-Impedance Adaptive Control for Intensive Contact Tasks

In tasks requiring intensive contact, such as assembly and drilling, unavoidable object interference leads to deviation from the contact task due to the interaction itself, which is normally accompanied by large contact forces and ultimately leads to task failure. To address these issues, a novel visual-impedance control framework based on human-guided operation is proposed. Specifically, we introduced an interaction model within the image space, which captures the dynamics of human–robot–environment interaction. Next, we characterize human manipulation skills via human-guided wrenches and project them as human-guided image features, unifying human manipulation wrenches and visual-impedance interactions. Ultimately, we develop an adaptive visual-impedance control strategy based on human-guided image features by utilizing the above framework. Impedance coefficients are updated by referencing human-guided features that implicitly characterize human operational skills in executing intensive contact tasks. The effectiveness of the proposed framework is validated through a series of intensive contact tasks.