Search papers, labs, and topics across Lattice.
This paper introduces a Dynamic Force Threshold Modulation Algorithm (DFTMA) for Hybrid Rigid-Flexible Dexterous Hands (HRSD-Hands) to improve safety in human-robot collaboration. DFTMA integrates ISO 13482 safety standards and Asimov's robot ethics using reinforcement learning-based tactile perception and a multi-dimensional safety threshold system. Experiments show DFTMA enhances safety in precision operations and collision scenarios while maintaining accuracy, offering a new approach to safe human-robot interaction.
By dynamically modulating force thresholds based on context and ethical principles, robots can now achieve safer and more reliable human-robot collaboration.
In response to the safety boundary challenges posed by human-robot collaboration in the context of embodied intelligence development, this study focuses on the safe operation of the Hybrid Rigid-Flexible Dexterous Hand (HRSD-Hand). The core innovation is the proposal of a Dynamic Force Threshold Modulation Algorithm (DFTMA) that integrates the ISO 13482 safety standard with Asimov's robot ethics. This algorithm combines reinforcement learning-based tactile perception, a multi-dimensional safety threshold system, and a context-aware conflict resolution mechanism to dynamically regulate maximum output force, stiffness, and response speed based on object attributes, environmental complexity, and human-robot distance. Experimental results demonstrate that DFTMA significantly enhances safety performance in precision operations and human-robot collision scenarios while maintaining accuracy, providing a new paradigm for safe human-robot collaboration and responsible AI applications.
