AUTONEWS

GPS data reveal why pedestrians in Phnom Penh rarely walk the shortest route

Cities across the Global South are urbanizing at pace, but their built environments for walking rarely resemble ideal, tidy, and well-regulated networks. Vehicles are parked on pedestrian sidewalks, and a seemingly direct route on a map may feel longer in practice. Moreover, many urban planning ideas and assumptions are based on research done on Global North cities like New York, which do not fully capture or reflect the nuances and lived experiences of the Global South.

Phnom Penh, the capital of Cambodia, is a case in point. Walkability in the city is complicated by uneven infrastructure, intense tropical heat, and widespread informal use of street space, from motorcycle parking on pavements to vendors occupying footpaths. Research Assistant Professor Samuel Chng from Singapore University of Technology and Design (SUTD) studied Phnom Penh's walking environment in his research paper "Walking the preferred path: exploring spatial-temporal variation in pedestrian route choice through revealed preferences in a developing city," published in Travel Behaviour and Society.

Together with his team at SUTD's Lee Kuan Yew Center for Innovative Cities, Asst Prof Chng analyzed over 6,000 walking trips drawn from anonymized mobile phone Global Positioning System (GPS) data. They combined the data with street view images and point-of-interest records to model how built-environment factors influence the routes pedestrians in Phnom Penh actually take.

Study sites in Phnom Penh, Cambodia representing the CBD, recreational, and commercial urban context Credit: SUTD

The team examined three contrasting districts: the Central Business District (CBD), the recreational area around the Independence Monument, and the commercial zone near the Russian Market. They also segmented trips by morning, midday, and evening periods. Using a Path Size Logit model, they measured how path efficiency, street vitality, and streetscape quality each contributed to route choice.

"We found that pedestrians do not simply choose the shortest path. They make trade-offs between efficiency, activity, and perceived comfort," said Asst Prof Chng. "This shows that walking is not just a calculation of distance, but a behavioral decision shaped by how a route feels and functions in context."

Shorter, simpler routes with fewer turns were consistently preferred, confirming efficiency as the dominant factor. However, the analysis revealed that each additional turn on a short walk was perceived as equivalent to roughly 40 meters of extra distance. This perception faded only on trips exceeding about 1.3 kilometers. Amenity density also mattered: every additional amenity along a route reduced perceived walking distance by about 10 meters.

One of the most counterintuitive findings concerned traffic. In the CBD, pedestrians gravitated towards busy roads and avoided crowded footpaths—a pattern that directly contradicts assumptions drawn from cities like New York or London.

Sample Mapillary images and selection of corresponding semantic segmentation Credit: SUTD

"In Phnom Penh, high-traffic streets are often the most structured and predictable parts of the network. They tend to be more regulated and less obstructed by informal uses," Asst Prof Chng explained. "Pedestrians are not choosing traffic itself, but clarity, predictability, and ease of movement. This reflects how people read their environment, where order signals usability."

Time of day introduced further variation. Morning walkers responded to greenery along their routes, midday walkers prioritized the shortest path to minimize heat exposure, and evening walkers favored amenity-rich, active streets that likely offered better lighting and a sense of safety. Greenery influenced route choice significantly during morning hours and in dense commercial areas, but had little effect elsewhere, suggesting that scattered vegetation in Phnom Penh is often too sparse to provide meaningful shade or thermal relief.

"It is about designing greenery that changes the walking experience in a meaningful way, especially in tropical climates," Asst Prof Chng noted.

The starkest result emerged when the team translated their model into perceived accessibility maps. In the CBD, the areas that residents could comfortably reach on foot shrank to just 37.7% of the objective 800-meter service area. The commercial zone fared better at 63.8%, while the recreational area sat in between at 52.4%.

"Cities may appear accessible on paper, but the reality is that many routes feel too complex, uncomfortable, or effortful," Asst Prof Chng said. "If this gap is ignored, cities risk overestimating how connected and inclusive they actually are."

The study acknowledges limitations: the anonymized GPS data cannot capture differences in age, gender or income, and the dataset covers a single month during the dry season. Future work could extend the framework across multiple seasons and incorporate socio-demographic attributes through data fusion approaches.

"The most actionable step is to prioritize direct, legible, and unobstructed pedestrian networks," Asst Prof Chng shared. "Before adding new features, cities should ensure that walking itself is straightforward and frictionless."

Provided by Singapore University of Technology and Design