At present, remote human-computer interaction terminals are mainly displayed in a stacked manner through video images and numerical icons. However, the lack of human factors design makes it difficult for monitors to understand information and has a high psychological load when dealing with faults, which affects the readability and accuracy of emergency response. Aiming to address the problem of insufficient human factors design in the remote terminal stacking interface information acquisition of unmanned electric tractors in facility greenhouses, an interface attention collection system was developed, and based on attention efficiency indicators such as reaction time and image recognition accuracy, attention efficiency experiments were conducted on the stacking interface when there were no faults and sudden single/two/three factor faults. The response law of attention efficiency to the stacking emergency interface was explored, and the evolution law of the distribution field of human factors attention on the interface was analyzed. Human factors engineering interface layout optimization design was carried out, the optimized distributed interface and centralized interface were verified through experiments, and the improvement effect of interface attention readability and accuracy was comprehensively evaluated. The research results showed that the average response time for single factor faults in stacked interfaces, distributed interfaces, and clustered interfaces were 1 096 ms, 1294 ms and 1 097 ms, respectively. The average response time for two factor faults was 1 123 ms, 1142 ms and 1293 ms, respectively. The fastest detection and warning response times for three factor faults were 820 ms, 1108 ms and 749 ms, respectively. The attention distribution fields exhibited distribution patterns of increasing modularity, increasing contrast, and centering, evolving towards change points and significant points, respectively. The interface optimization plan ultimately selected a clustered interface, which decreased the average response time of emergency response by 3.4% compared with that of a stacked interface, increased the average accuracy rate by 11.66% within 2 seconds and 34.94% within4 seconds, and reduced the psychological load on the monitor by 11.99%.