Foot traffic in Fukuoka shifts with weather and COVID rules, study finds

An AI system tracked how fast people walked during different times of day and during Japan’s COVID emergency periods.

Editorial illustration — not from the study.

Researchers built a computer-vision framework to extract individual walking paths from a year-long video dataset recorded by a fixed camera system in a high-traffic shopping district in Fukuoka, Japan. They used the system to measure walking speeds across four daily time slots—morning, afternoon, evening, and night—while also recording temperature and three coronavirus disease 2019 (COVID-19) State of Emergency periods.

They report that pedestrian responses were “strongly moderated” by walking purpose. Morning commuters walked faster and were more sensitive to temperature, while night-time pedestrians showed greater behavioral variability and responded more to emergency-period restrictions than to thermal conditions, alongside a demographic shift toward younger night-time walkers during emergency periods.

How temperature and COVID rules differed

The study reports that walking speed responses depended on both the environment (temperature) and policy (COVID-19 State of Emergency periods), but in a time- and purpose-dependent way. Morning commuters walked faster and were more sensitive to temperature. Night-time pedestrians showed greater behavioral variability and responded more to emergency-period restrictions than to temperature, and demographic analysis indicated a shift toward a younger nighttime population during emergency periods.

Why time-of-day matters for policy

The results suggest that averaging activity over a whole day can hide how different groups respond to different pressures. By showing how environmental and policy influences depend on time of day and walking purpose, the study provides evidence for more targeted evaluation of urban policies and for planning that accounts for real-world behavior.

Video-based tracking with place and data limits

This is an observational analysis of pedestrian movement in one high-traffic shopping district in Fukuoka, Japan, based on a year-long, fixed-camera video dataset. The study uses an AI system to detect and track pedestrians and then analyzes walking speed across time slots and policy periods. Limitations include that findings are tied to the specific location, camera setup, and the study’s categorization of walking purpose/time slots; the abstract does not specify how walking purpose was determined or how many pedestrians were analyzed.

A flying robot used touch to perch on unfamiliar structures

Researchers developed an aerial robot that uses touch to refine its position and perch on structures it has not encountered before. The system succeeded in more than 99% of simulated trials and completed 26 real-world perching trials despite corrupted position estimates.

4 hours ago·3 views

Abstract editorial illustration accompanying the article: Belgium’s low-carbon energy plans could cut climate pollution but raise other risks

Engineering & Technology

Belgium’s low-carbon energy plans could cut climate pollution but raise other risks

A modeling study of Belgium’s 2045 energy system found that all near-optimal low-carbon pathways met the climate target but exceeded limits for ecotoxicity and particulate matter under the most permissive assessment approach. The pathways had similar costs but differed in whether they relied mainly on electrification, hydrogen, renewable power capacity or a mix of strategies.

1 day ago·49 views

Engineering & Technology

A new traffic-planning method handles more controls across complex road networks

Researchers developed a way to calculate traffic sensitivities efficiently in road networks with different vehicle classes, merges, diverges and congestion spillback. Tests on a sample network found the approach was particularly efficient as the number of control parameters increased.

1 day ago·3 views