Intuitive interaction flow: A Dual-Loop Human-Machine Collaboration Task Allocation Model and an experimental study

Jiang Xu, Qiyang Miao, Ziyuan Huang, Yilin Lu, Lingyun Sun, Tianyang Yu, Jingru Pei, Qichao ZhaoarXiv 2024 · arXiv · OpenReview

Abstract

This study investigates the issue of task allocation in Human-Machine Collaboration (HMC) within the context of Industry 4.0. By integrating philosophical insights and cognitive science, it clearly defines two typical modes of human behavior in human-machine interaction(HMI): skill-based intuitive behavior and knowledge-based intellectual behavior. Building on this, the concept of 'intuitive interaction flow' is innovatively introduced by combining human intuition with machine humanoid intelligence, leading to the construction of a dual-loop HMC task allocation model. Through comparative experiments measuring electroencephalogram (EEG) and electromyogram (EMG) activities, distinct physiological patterns associated with these behavior modes are identified, providing a preliminary foundation for future adaptive HMC frameworks. This work offers a pathway for developing intelligent HMC systems that effectively integrate human intuition and machine intelligence in Industry 4.0.

Introduction

As Industry 4.0 advances, the paradigm of manufacturing is shifting from standalone automation to sophisticated Human-Machine Collaboration (HMC). However, this transition brings forth the "Automation Paradox": as machines become increasingly autonomous, human operators may lose the critical skills required to intervene during system failures. Traditional task allocation models often rely on static divisions of labor that fail to account for the dynamic quality of human engagement. To address this, researchers from Tongji and Zhejiang Universities propose a framework centered on "intuitive interaction flow," aiming to optimize collaboration by recognizing the cognitive state of the human operator.

The Dual-Loop Model

The core of this research is the Dual-Loop HMC task allocation model. Grounded in the Learning Intelligent Distribution Agent (LIDA) cognitive architecture and philosophical concepts from Heidegger and Bergson, the model distinguishes between two primary modes of human behavior:

  1. The Intuitive Loop: This loop characterizes expert performance where sensory input aligns with "habitual body schemas," allowing for fluid, automatic execution. In this state, the machine provides minimal auxiliary support to avoid disrupting the human's "flow."
  2. The Intellectual Loop: This loop is activated for novices or during complex, novel scenarios where sensory input does not match existing skills. This requires conscious, rule-based deliberation. Here, the machine provides maximum decision support and auxiliary collaboration.
The proposed Dual-Loop Human-Machine Collaboration (HMC) task allocation model.

By identifying these states, the system can adaptively adjust its level of intervention, stepping back when the human is in a state of flow and offering support when they struggle.

Experimental Methodology

To validate the model, the researchers conducted an experimental study involving a high-skill mechanical excavator task. The study compared 10 experts (average 15 years experience) with 10 novices (average 1.3 months experience). Participants performed two types of tasks: a routine "trenching" task and a complex "trench-finding" task requiring high judgment to avoid buried obstacles.

Physiological data was captured using an 8-channel EEG to monitor brain activity and surface electromyography (sEMG) to measure muscle activity in the dominant arm. A key metric analyzed was Corticomuscular Coherence (CMC), which quantifies the synchronization between the motor cortex and muscles, serving as an indicator of "perception-action coupling."

Results and Discussion

The experimental results revealed distinct physiological markers for different levels of expertise and task complexity:

  • Neural Efficiency in Experts: During complex tasks, experts exhibited significantly higher Alpha-band power in the EEG (specifically at C3, Cz, C4, and Fpz electrodes). In neuroergonomics, high alpha power is associated with "neural efficiency," where the brain suppresses distractions because the task has become second nature.
  • Cognitive Strain in Novices: Conversely, novices showed higher Beta-band activity across nearly all electrodes, indicating high alertness, selective attention, and a significantly higher cognitive workload.
  • CMC as an Expertise Fingerprint: The most significant finding was that experts showed much higher CMC in the beta and gamma bands during complex tasks. This suggests that for experts, the machine becomes a "transparent" extension of their own body—a state Heidegger refers to as "readiness-to-hand."
  • Task Complexity as a Differentiator: Interestingly, physiological differences were negligible during simple tasks. The "intuitive state" only becomes a measurable differentiator as task difficulty increases.

Conclusion

This study provides a preliminary foundation for adaptive HMC frameworks. By using physiological markers like Alpha/Beta Power Spectral Density (PSD) and CMC, machine learning algorithms can classify a user's cognitive state in real-time. While the study was limited by a small sample size (14 subjects after data cleaning) and environmental noise, it offers a clear pathway for developing intelligent systems that prevent skill degradation and optimize human-machine synergy in industrial environments. Future work will focus on real-time closed-loop implementation and long-term validation of the "Automation Paradox" mitigation.

Summary updated

Abstract

This study investigates the issue of task allocation in Human-Machine Collaboration (HMC) within the context of Industry 4.0. By integrating philosophical insights and cognitive science, it clearly defines two typical modes of human behavior in human-machine interaction(HMI): skill-based intuitive behavior and knowledge-based intellectual behavior. Building on this, the concept of 'intuitive interaction flow' is innovatively introduced by combining human intuition with machine humanoid intelligence, leading to the construction of a dual-loop HMC task allocation model. Through comparative experiments measuring electroencephalogram (EEG) and electromyogram (EMG) activities, distinct physiological patterns associated with these behavior modes are identified, providing a preliminary foundation for future adaptive HMC frameworks. This work offers a pathway for developing intelligent HMC systems that effectively integrate human intuition and machine intelligence in Industry 4.0.
BibTeX
@misc{xu2024intuitive,
  author = {Jiang Xu and Qiyang Miao and Ziyuan Huang and Yilin Lu and Lingyun Sun and Tianyang Yu and Jingru Pei and Qichao Zhao},
  title = {{Intuitive interaction flow: A Dual-Loop Human-Machine Collaboration Task Allocation Model and an experimental study}},
  year = {2024},
  eprint = {2410.07804},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2410.07804},
}
Figure