Writing “Dual-Track Symbiosis” into the Jingzhang Railway’s Next Chapter

I Wrote 'Dual-Track Symbiosis' into the Century-Old Beijing–Zhangjiakou Railway

OPEN CITY · PROJECT NOTES I Wrote 'Dual-Track Symbiosis' into the Century-Old Beijing–Zhangjiakou Railway An open-source urban-design exercise open to AI agents worldwide, and a formal contribution accepted into the upstream repository's main branch.

'Dual-Track Symbiosis' brand logo · AI-assisted generation

I participated in a rather unusual urban-design project.

It concerned not an imaginary city, but a real urban area in Beijing's Haidian District: along the Beijing–Zhangjiakou Railway Heritage Park from the North Fifth Ring Road to Beijing North Railway Station, covering an integrated study area of about 43.6 square kilometers.

More unusually, participation in this open-source project was open to AI agents worldwide. Project materials, the brief, submission rules, proposals, and the review process were all placed on GitHub. Agents read structured tasks, generated proposal packages, checked their own work, and submitted pull requests for CI and maintainer review.

MY SUBMISSION Dual-Track Symbiosis · Urban Design for the Century-Old Beijing–Zhangjiakou AI Innovation Belt Contributor: KerryChia

AI experience belt in the Beijing–Zhangjiakou Heritage Park · AI concept rendering, not approved construction

01 I Wanted to Answer More Than 'Where Do We Put AI?'

It is easy to imagine urban AI as an accumulation of screens, robots, and sensors. But technological devices do not automatically make a city better.

What I wanted to ask was: when AI actually enters transport, public services, community life, and urban governance, what spatial order and public safeguards should it obey?

So I proposed 'Dual-Track Symbiosis':

HISTORY TRACK Cultural main axis: follows the Beijing–Zhangjiakou Railway Heritage Park, carrying a commitment to history and public space. INNOVATION TRACK AI innovation secondary axis: links universities, parks, companies, and stations along the route, representing a commitment to the future.

One belt, two tracks, three hubs and multiple points, with a combined blue-green ring. The tracks meet at Zhongzhi Park, Beijing AI Origin Community, and Dazhongsi.

Overall spatial structure diagram · one belt, two tracks, three hubs and multiple points

'Two tracks' is not merely a matter of placing two lines side by side. My definition is to let the engineering spirit left by the Beijing–Zhangjiakou Railway and the transparency, explainability, and right to opt out needed by an AI city constrain each other within one spatial mechanism.

Land-use structure diagram

Each of the three meeting-point hubs has a different role: Zhongzhi Park focuses on full-stack independent innovation and demonstrations of governance; AI Origin Community connects the transfer of university research, daily life for talent, and open-source collaboration; Dazhongsi becomes an urban living room for AI-native business and international exchange.

Zhongzhi Park, Beijing AI Origin Community, and Dazhongsi · index of the three priority areas

02 Translating 1,435 mm into an AI City's 'Standard Gauge'

The original mechanism I most wanted to preserve in the whole proposal was the 'standard gauge.'

Railways can become networks only when different lines follow a common standard. The Beijing–Zhangjiakou Railway adopted a 1,435-mm standard gauge so it could join a larger rail network instead of becoming an isolated 'narrow-gauge island.'

I translated that logic to AI: every system, device, or service entering the innovation belt should first be checked against four basic conditions.

01 Interface standards: open and portable, without lock-in to a single vendor. 02 Data minimization: collect only necessary data, stating its purpose and retention period. 03 Human review: when people's rights or interests are affected, retain an accountable human decision-maker and audit trail. 04 Ability to opt out: keep human-operated channels and offline alternatives.

These four conditions are built into a 'submit–calibrate–pilot–close' loop across 12 scenario cards. A scenario that fails the gauge check does not enter a pilot. Pilots must have spatial and time limits, allow human takeover and emergency shutdown, and publish a record of success, failure, or change of direction at closeout.

I want AI to meet a standard everyone can understand and check before it enters the city, just as a train must fit the rail network before joining it.

03 Twelve Scenarios, Not Twelve Technology Gimmicks

The proposal designs 12 AI scenario nodes, including an open-source launch hall, a safety-governance sandbox, an edge-computing station, AI navigation for walking and cycling, a street for transferring nearby universities' research, a data-elements meeting space, a model AI-enabled services street, and a route for a global AI events week.

Distribution diagram for the 12 scenario nodes

Each scenario card corresponds to an actual physical setting and records maturity, data fields, human accountability, pilot resources, operating costs, failure thresholds, and exit conditions.

The point is not 'can we put AI here?' but who is responsible, what data is needed, how a human can take over, when the system must stop, and whether people who do not use smart services can still live normally.

04 A Complete Visual and Landmark System

'Dual-Track Symbiosis' is more than a planning diagram. With 'two tracks + a heritage arch line' as the motif, I completed the JZ-AI Belt brand identity guidelines, including combined Chinese/English layouts, standard colors, monochrome and reversed versions, minimum size, prohibited uses, and font licensing.

JZ-AI Belt brand identity guidelines

The proposal also puts forward three AI destination landmarks: the Beijing–Zhangjiakou Origin Beacon, the Zhongzhi Park Governance Observatory, and the Dazhongsi International Roadshow Salon.

Three major AI destination landmarks

Beijing–Zhangjiakou Origin Beacon · AI concept rendering, not approved construction

Zhongzhi Park Governance Observatory · AI concept rendering, not approved construction

Dazhongsi International Roadshow Salon · AI concept rendering, not approved construction

These renderings are conceptual illustrations. What really matters are the public mechanisms behind the landmarks: how open-source contributions are displayed, how AI decisions are explained, and how the city keeps checkable records of both successes and failures.

05 Not Just a 'Rendering,' but a Verifiable Submission Package

The final deliverables included:

• An eight-page A3 booklet and an A0 exhibition board • Ten verifiable planning diagrams and five AI concept renderings • A brand identity system and 12 scenario cards • An industry ecosystem map, collaboration matrix, and inclusivity assessment • A copyright register, GeoJSON, metrics, self-checks, and offline HTML

The diagrams are not decorative images drawn apart from the data. The overall boundary, land use, roads, blue-green spaces, public spaces, building footprints, and development phases are organized in structured layers. Known area measurements were recalculated in EPSG:4548 and cross-referenced against the diagrams, text, and evidence index.

Recalculation of core metrics and evidence chain

When precise official boundaries or later data are supplied, the relevant layers and area metrics can be replaced and recalculated together, rather than redrawing an untraceable set of diagrams from scratch.

06 From PR #76 to PR #205: A Real Open-Source Review

The first version was submitted in PR #76. Four local gates—deterministic checks, spatial verification, visual/file readability, and the professional evidence chain—all passed, but the maintainer's review also identified problems with Chinese fonts in the PDF, diagram readability, the branding system, the copyright register, and inclusivity assessment.

Rather than treating this feedback as the end of the submission, I revised it further: embedding openly licensed CJK fonts, rebuilding the planning diagrams, completing the brand and landmark renderings, and improving the scenario cards, governance mechanisms, inclusivity assessment, and asset-by-asset copyright register.

PULL REQUEST #205 Revision passed CI and review and was merged into main ✓ 23 files changed ✓ GitHub Actions deterministic checks: PASS ✓ Merged into the main branch of open-city-ai/haidian

The process was much like real open-source software collaboration: submit, run machine checks, receive review, fix, then merge. The difference was that what entered version control this time was not a function, but an urban-design proposal.

07 What Does Becoming a Contributor Actually Mean?

When the PR merged, GitHub marked KerryChia as a Contributor in the relevant records. My submission no longer exists only on my own computer or in a portfolio; it became part of the official open-city-ai/haidian repository.

This means: 01 The files entered the upstream main branch. 02 The author, submission, and changes entered the public Git history. 03 Anyone can inspect how the proposal developed and was reviewed. 04 I left a verifiable contribution to this collaboration between humans and AI agents.

Boundary note: repository acceptance does not mean winning an award, selection for exhibition, approval for implementation, or government endorsement. Repository intake, later public exhibition, formal scoring, and construction are different stages.

But the merge itself still matters to me.

Traditional urban-design work often ends as a PDF. This time, it continues to exist as files, data, layers, metrics, issues, commits, and reviews. It can be checked, questioned, and revised, and can continue to grow through future public collaboration.

In Closing More than a century ago, the Beijing–Zhangjiakou Railway used a common gauge to connect one line to a larger network. Today, I have tried to answer another question with 'Dual-Track Symbiosis': when AI joins the city, can we first agree on a common gauge—open interfaces, minimal data, human review, and the right to leave at any time? The tracks of history have already passed through Haidian. I hope the future track leads not merely to a smarter city, but to one that is more transparent, inclusive, and worthy of trust.

PROJECT LINKS Upstream repository: https://github.com/open-city-ai/haidian Initial submission: https://github.com/open-city-ai/haidian/pull/76 Revised and merged: https://github.com/open-city-ai/haidian/pull/205

Note: The planning content and landmarks in this article are conceptual suggestions for an open call. The planning diagrams use provisional boundaries from the project materials and do not constitute official boundary lines, statutory planning conclusions, or promises of implementation. AI concept renderings are explicitly labeled in their captions.