Jinmeng Rao, Ph.D.

At a glance

1Accomplishments

Geoprivacy: responsible use of location data (2019–present)

• Proposed LSTM-TrajGAN (GIScience 2021, 270+ citations), a deep generative model for privacy-preserving synthetic trajectories, released as open-source code (80+ stars, 30+ forks).
• Proposed CATS (IJGIS, 2023) for conditional adversarial trajectory synthesis in privacy-preserving data publication. Also developed a decentralized, privacy-preserving framework for location recommendation (Transactions in GIS, 2021) and VTSV, a privacy-preserving vehicle trajectory simulation platform (ACM SIGSPATIAL GeoAI 2021).
• Wrote a vision paper on privacy-preserving and secure GeoAI foundation models (ACM SIGSPATIAL 2023) and a book chapter on trajectory privacy protection approaches (2025).
• Dissertation Trajectory Privacy Protection with Geospatial Artificial Intelligence (2023), which received the 2024 AAG William L. Garrison Award (one recipient every two years).
• Uptake: LSTM-TrajGAN has become a reference model for synthetic trajectory generation. Other groups have extended it (to enforce differential privacy, IEEE Access 2026; for selective partial anonymization, IEEE Access 2023; with Transformer decoders, MLNLP 2022), built it into processing pipelines (IEEE Access 2023; vehicle-trajectory protection, IEEE Internet of Things Journal 2024), and re-implemented it as the state-of-the-art benchmark (PETS 2022 and 2024; Information Systems Research 2023). Industry researchers at Lenovo (FGCS 2024), State Grid (IEEE T-ITS 2022), Rakuten (IEEE BigData 2022), and ByteDance (IJGIS 2025) have drawn on the trajectory privacy work. Details are in the uptake table.

Public health: mobility analytics for COVID-19 response (2020–2021)

• Co-developed one of the first mobility-based social distancing maps and dashboards in the U.S. at the UW–Madison GeoDS Lab (SIGSPATIAL Special, 2020, 410+ citations). The tools used commercial location data (SafeGraph, Descartes Labs) and ArcGIS Dashboards, and were covered by local NBC, ABC, and FOX stations.
• Co-authored a multiscale dynamic human mobility flow dataset (Scientific Data, 380+ citations) and studies linking mobility and infection (JAMA Network Open, 270+ citations; PNAS, 190+ citations).
• This work has been cited in Nature (Chang et al., 2021; Buckee et al., 2021) and PNAS (Thomas et al., 2022). The research was supported in part by the NSF.
• Uptake: the open flow dataset (GitHub, 150+ stars) became a widely used input for mobility and epidemic modeling. For example, the Deep Gravity mobility-flow model co-developed by Amazon researchers (Nature Communications, 2021) was evaluated on it. Industry research teams at Microsoft Research (Nature, 2021), Google (Nature Communications, 2021), Meta, Tencent, Orange, IQVIA, and DiDi (WWW 2026) drew on these results in their own mobility and epidemic studies.

Community building: connecting industry, academia, and students (2019–present)

• Co-founded and directs GISphere (since 2019), a non-profit that helps students worldwide find and apply to GIS graduate programs, supporting the future geospatial workforce. Its survey of prospective GIS students is published in Transactions in GIS, 2023.
• Program Co-chair of the ACM SIGSPATIAL GeoIndustry Workshop for four consecutive editions (2023–2026), a venue built to reduce siloed efforts between geospatial industry and academia (details).
• Demo Track Co-chair of ACM SIGSPATIAL 2026, and organizer of the OASIS 2026 student challenge on open agents with spatial intelligence for social good (co-organized with researchers from Google, Esri, and universities), for which I secured Google DeepMind sponsorship for student participants. Also organizer of the UMFM 2026 workshop on urban mobility foundation models.
• Co-chair of the GeoAI and Deep Learning Symposium at the AAG Annual Meeting (2023–2026) and GeoAI session organizer every year since 2022, including sessions on geoprivacy and ethics in GeoAI (2023).
• Earlier service: Student Director of the AAG GIS Specialty Group (2021–2022), CPGIS committees (2022–2023), and volunteer for UN Mappers and the Wisconsin Land Information Association.

Broader GeoAI research and industry AI work

• GeoAI foundation models: co-authored Towards the Next Generation of Geospatial Artificial Intelligence (JAG, 2025, 220+ citations), drawn on by industry work at Esri (ISPRS J. Photogramm. Remote Sens., 2026), InferLink (KDD 2026), and Meituan (IJGIS, 2025); co-developed SRL (ACM SIGSPATIAL 2024) and TRAJGANR (arXiv, 2026).
• Agriculture and Earth observation (Mineral, Alphabet): multimodal LLMs and retrieval-augmented generation, including IM-RAG (SIGIR 2024); co-inventor of U.S. Patent 12,462,551 and two published U.S. patent applications on multimodal agricultural inference.
• Transportation: urban multiple route planning with reinforcement learning (IEEE T-ITS, 2021) and road network extraction from taxi GPS trajectories (Handbook of Big Geospatial Data).
• Google DeepMind: works on general AI research and geospatial reasoning research, and is a core contributor to Gemini 2.5, Gemini 3, and Gemini 4 Argon (post-training and evaluation).

2Impact on the geospatial industry

Why this work matters to the geospatial industry

• Geoprivacy for the location-data industry. Companies that collect or analyze location data (mobile apps, mobility platforms, connected vehicles) need to share and analyze trajectories without exposing individuals. LSTM-TrajGAN and CATS generate synthetic trajectories that keep the useful patterns of real data while protecting privacy. Industry researchers have drawn on this work in vehicle-network privacy (State Grid, IEEE T-ITS 2022), trajectory privacy models (Lenovo, FGCS 2024), and trajectory-user linking (Rakuten, IEEE BigData 2022).
• Commercial location data for public health. The COVID-19 maps and datasets turned commercial mobility data (SafeGraph, Descartes Labs) into open information through Esri ArcGIS Dashboards, used by the public, local media, and public health researchers. The open flow dataset became a benchmark for mobility models, including the Deep Gravity model co-developed by Amazon researchers, and the work is drawn on by researchers at Microsoft Research, Google, Meta, Tencent, Orange, IQVIA, and DiDi.
• Community that links industry and academia. GeoIndustry, the SIGSPATIAL 2026 demo track, and the OASIS challenge (co-organized with Google and Esri researchers) give industry and academic teams shared venues, while GISphere helps students worldwide enter GIS graduate programs and the geospatial workforce.

Global reach of the research

The map covers my geospatial and AI research, excluding large team technical reports such as Gemini. It shows where the authors of the 1,600+ distinct publications citing this work are based: 81 countries in total. Each citing publication is counted once, even if it cites several of my papers. Hover over a country to see its count.

Connecting industry and academia through GeoIndustry (2023–2026)

I have served as a Program Co-chair of the ACM SIGSPATIAL International Workshop on Spatial Big Data and AI for Industrial Applications (GeoIndustry) for four consecutive editions, from the 2nd edition (Hamburg, 2023) to the 5th (Riverside, 2026). The workshop's stated goal is to reduce siloed efforts between industry and academia. Industry brings geospatial data and problems at scale, and academia brings new methods. The workshop gives both sides a recurring venue to exchange ideas through invited and regular talks.

• The program chairs and committee include researchers from Amazon, Google, Google DeepMind, Oracle, and PAII, together with faculty from the University of Wisconsin–Madison, the University of Maryland, UT Austin, Florida State University, and the University of Georgia.
• Workshop papers are published by ACM in the GeoIndustry proceedings.
• Editions: 2023 · 2024 · 2025 · 2026

Other industry–academia events at ACM SIGSPATIAL 2026

• Demo Track Co-chair, ACM SIGSPATIAL 2026. The demo track is where academic and industry teams present working geospatial systems (organizing committee).
• Organizer, OASIS 2026. This is the first ACM SIGSPATIAL student challenge on open agents with spatial intelligence for social good. It is co-organized by researchers from Google, Esri, and several universities. I secured Google DeepMind sponsorship to support student participants.
• Organizer, UMFM 2026. A workshop on urban mobility foundation models.

How others have extended, adopted, and used the work

Beyond citation, other research groups have built new methods on my work, adopted it as the benchmark to beat, and used the released data and code in their own systems and studies. Semantic Scholar classifies 22 citations of LSTM-TrajGAN by other authors as “highly influential”, meaning the citing paper relies substantially on it. Selected examples:

Sources: Semantic Scholar citation contexts and GitHub, October 2026. Descriptions follow the citing papers' own wording.

Industry research drawing on the work

Based on OpenAlex affiliation data (October 2026, self-citations excluded), about 60 publications co-authored by researchers at 34 companies across the United States, Europe, and Asia cite my geospatial research. Each example below links to the industry paper and to the specific work of mine that it cites. Where the citation context is available (Semantic Scholar), it is summarized or quoted.

Other companies in the data include Baidu, Alibaba, Nokia, Canon, Ipsos, Merck, Nommon, LocationMind, IBI Group, and Tenaga Nasional Berhad.

Engagement with geospatial practitioners

• Presented Mapping Mobility Changes During COVID-19 Using ArcGIS Dashboards and Mobile Experience Builder at the Esri Education sessions, AAG 2023. Also had an extended abstract on federated learning for location data accepted at the Esri User Conference (2020).

Media coverage

• The UW–Madison COVID-19 social distancing maps and dashboards, which I co-developed, were featured by local television news: NBC15, WISN 12 (ABC), and FOX47.
• The dashboard was hosted publicly by the UW–Madison Department of Geography for use by the public, media, and public-health researchers.

3Aspiration

My goal is to help the geospatial industry use location data and AI in ways that are useful, responsible, and open. Over the next several years, I hope to:

• Make privacy protection a standard part of geospatial AI. Methods for protecting location data should be practical enough for location-data providers and mobility platforms to adopt by default, and should extend to geospatial foundation models and AI agents.
• Apply location data and spatial reasoning to public health and social good. This includes my geospatial reasoning research at Google DeepMind and benchmarks and challenges, such as OASIS, that test whether AI models and agents can reason reliably about places, movement, and environmental change.
• Keep building the community. Continue GeoIndustry and SIGSPATIAL venues where industry and academia work on shared problems, and continue supporting students entering the geospatial field through GISphere.

Recognition

• 2024William L. Garrison Award for Best Dissertation in Computational Geography, American Association of Geographers (AAG); awarded to one recipient every two years
• 2024Spotlight Paper, AAAI 2024 Workshop on Responsible Language Models
• 2022SRGC Conference Presentation Award; Trewartha/Odebolt Conference Travel Award, University of Wisconsin–Madison
• 2019Best Poster Paper, ACM SIGSPATIAL RAAS Workshop
• 2015, 2017National Scholarship, Ministry of Education, China

Service to the geospatial community

Selected invited talks

• 2023Mapping Mobility Changes During COVID-19 Using ArcGIS Dashboards and Mobile Experience Builder, Esri Education sessions, AAG Annual Meeting
• 2023Privacy and Security Issues in Large-Language-Model-powered Geospatial Applications, Geography According to ChatGPT, University of California, Santa Barbara
• 2023Privacy-Preserving Location Recommendation via Decentralized Collaborative Learning, Built Environment Talk Series, Texas A&M University
• 2023Deep Reinforcement Learning for Driving Trajectory Simulation, Reinforcement Learning Seminar, Anhui University
• 2023Geographic Visualization and ArcGIS Dashboards, Geovisualization and Spatial Analysis Seminar, University of Arkansas
• 2022Trajectory Privacy Protection with Geospatial Artificial Intelligence, UW Roundtable on the Frontiers of Geospatial Artificial Intelligence, University of Washington

Supporting links

• Google Scholar profile
• Full publication list
• Professional service
• Awards
• ACM SIGSPATIAL 2026 organizing committee
• GeoIndustry 2026 workshop
• OASIS 2026 student challenge
• UMFM 2026 workshop
• COVID-19 mobility dashboard (UW–Madison)
• GISphere