Based on the high learning cost of the past walking simulation, this paper develops a prototype of interactive multi-agent walking simulation based on the game software unity 3D. The establishment of the workflow can facilitate the design to carry out simple walking experiments, which is convenient for future design and research.
PROTOTYPE DEVELOPMENT
原型机有四个组件：hierarchy window, the scene view, the inspector window, and the game view
hierarchy window
该窗口提供构件的整理和分级功能，不仅仅包括游戏中的物理组件，还包括运行逻辑和脚本。
scene view
在这个窗口可以定位和移动游戏对象
inspector window
检视窗口可以了解游戏对象的属性和参数，是自定义脚本的重要窗口。
game view
这个窗口是可以完成脚本运行的窗口
the prototype has four components: hierarchy window, the scene view, the inspector window, and the game view
hierarchy window
This window provides the ability to organize and rank components, not only the physical components of the game, but also the running logic and scripts.
scene view
In this window you can locate and move game objects
inspector window
The view window allows you to understand the properties and parameters of the game object and is an important window for custom scripts.
game view
This window is the window in which the script can be executed
原型机有三个原则：
第一
监视窗口的编程要尽可能简单易懂，不超过使用者的理解。
第二
UI要尽可能减少步骤，只保留增加减少对象和编写脚本这两大功能。
第三
仿真模块的图形一定要简单易懂，大家不用怎么反应就能够理解其意义
The prototype has three principles:
first
The programming of the monitoring window should be as simple as possible and not exceed the user's understanding.
second
The UI should have as few steps as possible, keeping only the two main functions of adding and removing objects and writing scripts.
third
The graphics of the simulation module must be simple and easy to understand, and everyone can understand its meaning without any reaction
原型机根据这三个原则可以得到工作流程：
第一
通过代理人生成器生成出特定数量的代理人，代理人生成器决定了代理人的最大数量。并能够通过生成脚本和代理人之间进行绑定。
第二
代理人的运行规则是通过设置满意和不满意的值来决定的，满意度由距离、速度、直线率、个人空间四个方面决定。
第三
在街道上可以放置传感器，检测代理人数据，可以设定传感器的物理监测范围，这样可以对整个街道的形态和人流进行科学地分析。
第四
在实验开始后，还可以实时地进行修正，既可以改变代理人生成器也可以改变代理人运行脚本。
第五
利用Constructive solid geometry可以改变街道的形态和大小，使我们可以更加容易模拟。
The prototype can get the workflow according to these three principles:
first
A specific number of agents are generated through an agent generator, which determines the maximum number of agents. And the ability to bind between the generated script and the agent.
second
The operating rules of the agent are determined by setting the values of satisfaction and dissatisfaction, and the satisfaction is determined by four aspects: distance, speed, straight-line rate and personal space.
third
Sensors can be placed on the street to detect agent data, and the physical monitoring range of the sensor can be set, so that the shape of the entire street and the flow of people can be scientifically analyzed.
Number four
After the experiment starts, it can also be modified in real time, either by changing the agent generator or by changing the agent run script.
fifth
By using Constructive solid geometry, we can change the shape and size of the street so that we can simulate it more easily.
实验案例
Experimental case
实验案例列举的是一个2020国际会议上的例子，目的是在covid-19的情境下做出更适宜人步行的社区。实验的流程大致是选择15人的实验人员，受实验人员预下载提供的软件并且提供例子作为训练样本。通过两个小时的训练对可步行性和实验案例进行讲解，使得受实验人员得到充分的训练。将受实验人员分为三组，按照不同的方式选择自己偏好的进行设计，探索不同的城市形态。然后通过这个插件不断实验街道形态和行人可步行性。
案例1 ADELAIDE CBD
该组对阿德莱德市中心的街道进行可步行设计，运用多种改善街道可步行性的方式。包括不断扩大的小路,提出通过底层行人连接,还有多增加坡道和倾斜的风景地带。这个案例研究演示了如何灵活的使用该设计框架可以支持快速迭代的设计在一个紧迫的时间表在务实的设计实践。
案例2 A RESIDENTIAL COMMUNITY NEAR THAMES RIVER, LONDON
该组有比较多的大数据处理基础，因此选择对伦敦东部住宅区的人流进行尽可能地分析，他们选择运用twitter地理空间数据作为人流的参考标准，将该插件的功能发挥地最为科学。
案例3 SHANGHAI NORTH BUND
这个项目中，其运用该插件对上海北外滩城市形态的变化如何影响行人的行为。在所有模拟,代理人数量和分布故意保持不变,但平均旅行时间和距离随着渗透率的增加而减少。
The test case is an example from a 2020 international conference aimed at making communities more walkable in the context of covid-19. The process of the experiment is roughly to select 15 experimenters, who pre-download the provided software and provide examples as training samples. The walkability and experimental cases were explained through two hours of training, so that the subjects were fully trained. The subjects were divided into three groups and selected their own preferences for design in different ways to explore different urban forms. Then, through this plugin, we continue to experiment with street shape and pedestrian walkability.
Case 1 ADELAIDE CBD
The team designed streets in Adelaide's city centre for walkability, using a variety of ways to improve street walkability. These include expanding paths, proposed pedestrian connections through the ground floor, and increased ramps and sloping scenic areas. This case study demonstrates how flexible use of the design framework can support rapid iterative design on a tight schedule in pragmatic design practices.
Case 2: A RESIDENTIAL COMMUNITY NEAR THAMES RIVER, LONDON
This group has a lot of big data processing basis, so they choose to analyze the flow of people in east London residential areas as much as possible. They choose to use twitter geospatial data as the reference standard for flow, and make the function of this plug-in play the most scientific.
Case 3 SHANGHAI NORTH BUND
In this project, it uses the plug-in to investigate how the change of urban form in the North Bund of Shanghai affects pedestrian behavior. The number and distribution of agents remained deliberately constant across all simulations, but the average travel time and distance decreased as penetration increased.
