Innovative Solutions to Address Poverty-Related Transport and Mobility Challenges

Satellite imagery can be harnessed to monitor road quality. Photo credit: ADB.

Share on:           

Published:

Leverage machine learning and satellite imagery for informed resource allocation to enhance road quality and address development challenges.

Introduction

Geography significantly influences socioeconomic mobility, with rural residents facing heightened disadvantages due to poor infrastructure. The recent Asian Development Bank (ADB) Key Indicators report reveals that approximately three out of every five people living in extreme poverty in developing Asia are found in rural areas. Empirical evidence underscores the enduring nature of chronic poverty in remote rural settings of low-income economies, where various socioeconomic disadvantages persists, including limited access to essential services like safe drinking water and sanitation (see Figure 1).

Figure 1: Proportion of Rural Population with Access to Safely Managed Drinking Water and Sanitation Services

Lao PDR = Lao People's Democratic Republic, PRC = People's Republic of China.
Source: Asian Development Bank estimates using data presented in Table 1.6.1 of Key Indicators for Asia and the Pacific 2023.

The poverty premium, wherein the poor pay more for certain goods and services, presents a substantial socioeconomic challenge in rural areas. Limited transportation options restricts access to affordable commodities, emphasizing the connection between the poverty premium and poor mobility.  Inadequate transport infrastructure increases long-term costs, including monetary expenses from regular commuting and non-monetary costs related to travel time and convenience. This results in reduced access to essential social services, including health care, education, and employment, intensifying social losses through limited participation in socioeconomic activities. Gender disparities are pronounced, with poor women often experiencing less mobility than their male counterparts.  Moreover, the lack of mobility options exposes the poor to increased risks, as seen during the height of the COVID-19 pandemic when they faced exposure due to a lack of alternatives to address their financial challenges.

How can improved transportation enhance living conditions?

Numerous studies, including those of  Seetanah et al. (2009), Dawkins et al. (2015), and Cai et al. (2023), have already demonstrated the promising impact of improved transport infrastructure in alleviating living conditions for the poor. Benefit-to-cost ratio analyses of public investments in transportation, particularly road development, reveal positive results for poverty alleviation in both rural and urban areas in certain developing economies (Gibson et al., 2003; Dillon et al., 2011; Fan, 2008). Moreover, transportation-related developments bring social benefits, especially for households located along improved road corridors, providing better access to social amenities like education, health care, and markets (Asomani-Boateng et al., 2015). The cited studies, based on empirical and simulated results, suggest that improving transportation infrastructure has  the positive effect of opening up channels for more resources and opportunities to flow into poorer communities and economically disadvantaged areas, eventually leading to the reduction or elimination of the poverty premium among poorer households.

In terms of road quality, the figure below reveals that developing Asian economies scored an average of only 4 out of 7, with many registering around 3. This poses an important development challenge for governments and practitioners working to minimize poverty premium, especially for those in remote rural areas.  Under SDG 9.1, focusing on "developing quality, reliable, sustainable and resilient infrastructure,”  the Rural Accessibility Index (RAI) serves as a key progress indicator. It measures the proportion of rural population within 2 km from an all-season road.

Figure 2: Quality of Roads in Selected Economies of Developing Asia

OECD = Organisation for Economic Co-operation and Development, PRC = People's Republic of China.
Source: Asian Development Bank estimates using data from the World Economic Forum. Global Competitiveness Report 2015-2016.

What approach can be employed to identify gaps in road infrastructure?

The Rural Access Index is useful in identifying gaps in road infrastructure in rural areas. To compile the index, three data requirements are necessary: geospatially tagged data on population, road networks, and all-season roads. However, gathering information on all-season roads faces challenges, as the definition varies across countries. Some assess based on visual conditions, average vehicle speed, or road roughness making intercountry comparisons difficult (Workman and McPherson 2021). Data collection, especially using the international roughness index (IRI) is challenging and costly, as it involves road quality surveys with sensors on a car wheel traveling along a given road segment at a predetermined speed.  

How can satellite imagery and machine learning technologies be harnessed for road quality monitoring?

An ongoing ADB study, conducted in collaboration with other development partners and funded by the Japan Fund for Prosperous and Resilient Asia and the Pacific, suggests that leveraging innovative data sources and employing machine learning enables an alternative, cost-effective, and efficient method for collecting data on road quality. For instance, machine learning algorithms assess road quality data using medium-resolution satellite imagery, addressing budget constraints. This approach fills the gap in collecting data for SDG 9.1.1 or the Rural Accessibility Index, which emphasizes the importance of rural areas in fostering development.

Computer vision techniques are capitalized for this methodology. Satellite imagery trains the algorithm for feature extraction and road roughness prediction. To manage budget requirements, medium-resolution satellite imagery is used. Images are downloaded via the Google Earth Engine API. To predict road quality based on the downloaded images, three models were developed: the Convolutional Neural Network (CNN) model, the Tabular Model, and the Combined Model.

CNNs excel in image recognition. They utilize a transfer learning approach, where a pre-trained architecture is used for image classification. The Convolutional Neural Network model developed for this study is trained and tested using 80% and 20% of the downloaded satellite images with existing road quality data.

The tabular model, with a three-layer and a five-layer neural network, does not use  transfer learning. lt incorporates Moderate Resolution Imaging Spectroradiometer (MODIS) temperature data, NASA Shuttle Radar Topography Mission (STRM) elevation data, European Centre for Medium-Range Weather Forecasts (ECMWF) precipitation data, and Worldpop population data.

To maximize benefits, a combined model processes road visual and tabular data simultaneously, classifying road quality using integrated outputs from the two models.

How accurate are the predictions of the models?

To assess the accuracy of the model, both 4-class (good, fair, poor, bad) and binary (good, bad) classifications of road quality were examined. Results show higher accuracy for binary classification, reaching 75%, compared to the 4-class classification with up to 60% accuracy. While the proposed methodology may not entirely replace conventional methods, it demonstrates the potential for the preliminary identification of road segments needing maintenance.

Can the accuracy of the models be improved?

Enhancing model accuracy involves considering adjustments. For the Convolutional Neural Network model, improved accuracy may result from using high-resolution satellite imagery or implementing super-resolution techniques. Alternative architectures like EfficientNet can be explored for the CNN model. For the Tabular model, accuracy may be influenced by the tabular data incorporated into the model.

Conclusion

Navigating complex development challenges, such as reducing the poverty premium, especially when this premium is attributed to high transportation costs resulting from infrastructure quality, necessitates the utilization of innovative data sources and analytical techniques. These include remote sensing data and advanced machine learning techniques, capable of offering timely and detailed insights into subjects like road quality. This initiative goes beyond mere data gathering; it represents a transformative strategy aimed at deriving practical insights to guide the allocation of resources where they are most needed to improve road quality.

Resources

A. Dillon et al. 2011. Estimating the Impact of Rural Investments in Nepal. Food Policy. 36 (2). pp 250-258.

Asian Development Bank (ADB). 2023. Key Indicators for Asia and the Pacific 2023. Manila.

A. Thegeya et al. 2022. Application of Machine Learning Algorithms on Satellite Imagery for Road Quality Monitoring: An Alternative Approach to Road Quality Surveys. ADB Economics Working Paper Series: 675. Manila.

B. Seetanah et al. 2009. Does infrastructure alleviate poverty in developing countries? International Journal of Applied Econometrics and Quantitative Studies. 6 (2). pp. 31-36.

C. Dawkins et al. 2015. Vehicle Access and Exposure to Neighborhood Poverty: Evidence from the Moving to Opportunity Program. Journal of Regional Science. 55 (5). pp. 687-707.

D. Hernandez. 2018. Uneven Mobilities, Uneven Opportunities: Social Distribution of Public Transport Accessibility to Jobs and Education in Montevideo. Journal of Transport Geography. 67. pp. 119-125.

F. Agbenyo et al. 2017. Accessibility Mapping of Health Facilities in Rural Ghana. Journal of Transport & Health. 6. pp. 73-83.

J. Cai et al. 2023. Transport Accessibility and Poverty Alleviation in Guizhou Province of China: Spatiotemporal Pattern and Impact Analysis. Sustainability. 15 (4). P. 3143.

J. Gibson and S. Rozelle. 2003. Poverty and Access to Roads in Papua New Guinea. Economic Development and Cultural Change. 52 (1). pp. 159-185.

K. Lucas. 2011. Making the Connections between Transport Disadvantage and the Social Exclusion of Low Income Populations in the Tshwane Region of South Africa. Journal of Transport Geography. 19 (6). pp. 1320-1334.

M. Adeel et al. 2016. Transportation Disadvantage and Activity Participation in the cities of Rawalpindi and Islamabad, Pakistan. Transport Policy. 47. pp 1-12.

R. Asomani-Boateng et al. 2015. Assessing the Socio-Economic Impacts of Rural Road Improvements in Ghana: A Case Study of Transport Sector Program Support (II). Case Studies on Transport Policy. 3 (4). pp 355-366.

S. Boonyabancha et al. 2019. How the Urban Poor Define and Measure Food Security in Cambodia and Nepal. Environment and Urbanization. 31 (2). pp. 517–532.

S. Fan and C. Chan-Kang. 2008. Regional Road Development, Rural and Urban Poverty: Evidence from China. Transport Policy. 15 (5). pp. 305-314.

Dennis Dizon
Consultant, Economic Research and Development Impact Department, Asian Development Bank

Dennis is an economics and statistics specialist at the Asian Development Bank. He is part of the team compiling various indicators for the ADB Key Indicators for the Asia and the Pacific report. Prior to joining ADB, he earned his bachelor’s degree in Geography and master’s degree in Statistics at the University of the Philippines-Diliman, and a master’s degree in Cartography at the Technical University of Munich.

Jayzon Mag-atas
Consultant, Economic Research and Development Impact Department, Asian Development Bank

Jayzon is a transport specialist at the Asian Development Bank. He is part of the team studying the effective use of high-level technologies for planning high-quality infrastructure investments, particularly in transport.. Prior to joining ADB, he earned his bachelor’s degree in Civil Engineering from the University of the Philippines-Los Baños.

Arturo Martinez, Jr.
Statistician, Economic Research and Development Impact Department, Asian Development Bank

Art Martinez works on Sustainable Development Goals indicator compilation, particularly poverty statistics and big data analytics. Prior to joining ADB, he was a research fellow at the University of Queensland where he also got his doctorate in Social Statistics.

Asian Development Bank (ADB)

The Asian Development Bank is committed to achieving a prosperous, inclusive, resilient, and sustainable Asia and the Pacific, while sustaining its efforts to eradicate extreme poverty. Established in 1966, it is owned by 68 members—49 from the region. Its main instruments for helping its developing member countries are policy dialogue, loans, equity investments, guarantees, grants, and technical assistance.

Follow Asian Development Bank (ADB) on
Leave your question or comment in the section below:
Disclaimer

The views expressed on this website are those of the authors and do not necessarily reflect the views and policies of the Asian Development Bank (ADB) or its Board of Governors or the governments they represent. ADB does not guarantee the accuracy of the data included in this publication and accepts no responsibility for any consequence of their use. By making any designation of or reference to a particular territory or geographic area, or by using the term “country” in this document, ADB does not intend to make any judgments as to the legal or other status of any territory or area.