Abstract

This paper introduces the Agent-Based Spatial Programming Design System (ABSPD), a framework that combines spatial adjacency analysis, agent-based modeling, and Virtual Reality (VR) to improve community participatory design. Unlike traditional “survey + 2D mapping” methods, ABSPD allows participants to build their own layouts directly in VR. These individual layouts are then merged into a shared 3D plan using constraint-based optimization. A case study with 48 participants shows that ABSPD increases engagement, transparency, and collaboration, while also producing layouts with higher adjacency satisfaction, fewer iterations, and better scores in relevance, realism, and satisfaction. The results suggest that ABSPD can connect qualitative community preferences with quantitative spatial logic, offering a more inclusive and data-driven pathway for participatory design.

Keywords

Spatial Programming Design, Virtual Reality, Spatial Adjacency, Participatory Design, Agent-Based Modeling

1. Introduction

Spatial adjacency (SA) analysis is a rule-based method for describing how spaces relate to each other and for optimizing layouts based on proximity requirements. Traditionally, architects use tools like bubble diagrams, adjacency matrices, or space syntax to guide design, focusing on circulation, efficiency, and usability. However, in participatory settings, these tools are rarely used effectively. Common approaches—such as surveys, interviews, and 2D mapping workshops—focus more on people’s preferences than on the underlying spatial logic. Moreover, existing methods often represent space in simplified units that do not reflect real experience, and adjacency frameworks usually rely on fixed, pre-defined spaces. This mismatch highlights a gap between professional design tools and community members’ lived perspectives.

This study explores how combining SA analysis with immersive VR can improve participatory design. Our goals are:

To make spatial relationships accessible and editable in VR, lowering the knowledge barrier for community participants.

To compare the outcomes with conventional “survey + 2D mapping” approaches.

We present the ABSPD system, which bridges the gap between professional design tools and public input. ABSPD allows participants to manipulate spatial relationships in VR, while the system combines these inputs into a consensus 3D scheme using agent-based modeling and constraint optimization. The result is a process that is more inclusive, transparent, and effective.

2. ABSPD and Related Work

ABSPD integrates agent-based modeling with spatial programming principles to support participatory design. In this system:

Each design element (e.g., seating, kiosks, facilities) is treated as an “agent” with attributes, roles, and adjacency rules.

Agents follow both design rules (e.g., clearance, connectivity) and user-defined preferences (e.g., “must be near,” “avoid”).

The system uses multi-agent interactions and optimization to generate 3D layouts that balance personal inputs with collective needs.

Unlike traditional adjacency methods based on static diagrams, ABSPD uses VR visualization so participants can view, check, and edit spatial relationships directly. This makes the process more transparent and accessible, encouraging active collaboration.

3. Method

ABSPD works through a six-stage process:

Module Library: Collect typical urban/architectural modules (e.g., seating, kiosks) and standardize them with labels and geometry.

VR Authoring: Participants use VR to place modules and mark spatial preferences (“must be near,” “should be near,” “avoid”). The system also detects adjacency from placement.

Personal Adjacency Extraction: Each layout is converted into a personal adjacency matrix, reflecting explicit and implicit preferences.

Consensus Graph: Personal matrices are merged into a weighted consensus adjacency graph.

3D Optimization: The system computes 3D layouts that respect adjacency, avoid overlap, and ensure circulation, refined with optimization algorithms.

Evaluation: Both objective (adjacency satisfaction, error rates) and subjective (relevance, realism, satisfaction) metrics are used to compare with traditional methods.

4. Case Study

We tested ABSPD with three groups: delivery workers, ride-share drivers, and simulation enthusiasts, involving 48 participants in total. Half used the traditional survey + 2D mapping method, while the other half used ABSPD. In the ABSPD condition, each participant created a personal VR layout, which was then aggregated into three consensus 3D designs.

5. Results and Discussion

The study found that ABSPD outperformed traditional methods in four areas:

Engagement: Participants were more motivated and expressive in VR.

Transparency: Spatial relationships were visible and easy to understand.

Collaboration: Shared VR environments reduced negotiation time.

Outcome Quality: ABSPD produced layouts with higher adjacency satisfaction, fewer errors, and better subjective ratings.

The weighting system also helped balance different roles’ needs and made trade-offs more transparent to participants.

6. Conclusion and Future Work

This study shows that ABSPD, by integrating spatial adjacency analysis, VR, and agent-based modeling, can significantly enhance community participatory design. While this study focused on short-term sessions with a relatively small sample, future work should explore larger-scale, long-term adoption and integration with live civic data or digital twins.

Overall, ABSPD not only improves participation and layout quality but also bridges the gap between professional spatial logic and community preferences. This makes participatory design more inclusive, transparent, and effective.