《景觀設計學》2018年第2期

人工智能時代的設計

Design in the Age of Artificial Intelligence

作者：尼爾·林奇 Neil LEACH

摘要

人工智能在我們的生活中扮演著越來越重要的角色，但它會對設計專業帶來何種影響？它會形成一種新的設計風格嗎？或者，人工智能僅能輔助設計，還是會顛覆設計？通過定義不同形式的人工智能并展示三位設計師及團隊的相關設計成果，本文試圖對人工智能的潛力發起討論。作者認為，人工智能不會創造新的設計風格；但其對設計過程影響深遠。這或許將迫使我們反思許多廣為設計界所接受的真理信條。某些傳統術語（例如，“設計”這一概念以及對于“設計”這一行為的認識）可能會被“搜索”和“結果”等一批新術語所取代。更重要的是，對于“藝術天賦”的迷思可能也會被打破。本文并非意在指出人工智能強烈的“人工性”，而是旨在點明我們對人類智能的理解已深陷誤區。

關鍵詞

人工智能；人類智能；功能可見性；搜索；結果；藝術天賦

Abstract 

Artificial Intelligence is playing an ever-increasing role in our lives, but what impacts might it have on the design professions? Will it introduce a new style? Or will it just help to improve the design process? Or will it have a completely different impact? This article attempts to offer an overview of the potential of Artificial Intelligence, defining the different forms of Artificial Intelligence, and illustrating its argument with the work of three designers. It argues that Artificial Intelligence will not generate a new style. However, it will have a radical input on the process of designing. It is also likely to force us to call into question many accepted beliefs within the design community. Certain traditional terms, such as the concept of a “design” and the very notion of “designing,” are likely to be replaced by a new lexicon that includes terms such as “searching” and “outcomes.” But, more importantly, the whole myth of the “artistic genius” is likely to be also called into question. What Artificial Intelligence tells us, as the paper concludes, is not how “artificial” Artificial Intelligence is, but rather how misguided our own understanding of human intelligence has been. 

Key words

Artificial Intelligence; Human Intelligence; Affordances; Searching; Outcomes; Artistic Genius

基于街道圖像與深度學習的城市景觀研究

Using Street-Level Images and Deep Learning for Urban Landscape Studies

作者：李小江，蔡洋，卡洛·拉蒂 Xiaojiang LI, Bill Yang CAI, Carlo RATTI

摘要

城市街道不僅是人類活動的集聚地，也是居民與城市建成環境發生社會交互的主要界面。因此，加深對于城市街道景觀的了解在城市研究工作中至關重要。街道圖像獲取性的大大提高為城市景觀研究提供了新的機遇，也提高了街道景觀研究與分析的準確性與多樣性。本研究基于街道圖像，呈現了新近研發的深度卷積神經網絡在景觀分析中的應用。利用經過訓練的深度卷積神經網絡模型，我們能夠準確地從街道圖像中識別出不同的城市特征。根據圖像分割技術處理結果，我們進一步測算出了馬薩諸塞州劍橋市的街道綠化空間分布情況，并對街谷開闊程度進行了量化分析。諸如上述人工智能與大規模采集的街道圖像的結合，將為世界范圍內的城市景觀研究提供全新的視角。

關鍵詞

卷積神經網絡；城市街道；人工智能；機器學習；圖像分割

Abstract 

Streets are a focal point of human activities and a major interface of the social interaction between urban dwellers and urban built environment. A better understanding of the urban landscapes along streets is thus important in urban studies. The increasing availability of street-level images provides new opportunities for urban landscape studies to study and analyze streetscapes at a fine level and from a different perspective. In this study, we presented an application of a recently developed Deep Convolutional Neural Network on landscape analysis based on street-level images. Different urban features were identified from street-level images accurately using a trained Deep Convolutional Neural Network model. Based on the image segmentation results, we further measured the spatial distribution of the street greenery and quantitatively analyzed the openness of street canyons in Cambridge, Massachusetts. The proposed combination of Artificial Intelligence and the massively collected street-level images provides a new sight for urban landscape studies for cities around the world. 

Key words

Convolutional Neural Network; Urban Street; Artificial Intelligence; Machine Learning; Image Segmentation

基于智能交互的景觀體驗增強設計

Application of Intelligent-Interaction-Based Landscape Experience Augmentation

作者：曹靜，何汀瀅，陳箏 Jing CAO, Tingying HE, Zheng CHEN

摘要

自20世紀90年代中后期起，人工智能和可穿戴交互技術不斷革新我們和環境交互的方式，而在未來，這種影響將會愈加深遠。可穿戴設備和移動終端主要從感知、理解和控制三個方面增強人對于環境空間的體驗。本文圍繞智能感官增強、智能認知增強，以及智能反饋增強三個方面，介紹了美國華盛頓紀念碑現場增強音樂項目、由麻省理工學院媒體實驗室城市科學研究團隊的學者為非專業用戶提供智能決策輔助的“城市視景”項目，以及基于地理位置服務的“自然控”自然社區營造實景增強現實游戲應用程序設計等若干創新研究和實踐探索，并展望了未來智能交互背景下空間環境規劃設計領域可能面臨的機遇和挑戰。

關鍵詞

智能交互景觀；可穿戴設備；體驗增強；增強現實

Abstract 

Since the mid- to late 1990s, both Artificial Intelligence and wearable interaction techniques have dramatically changed the way how humans interact with the exterior environment, as such influence will be greater and greater over time. Wearable devices and mobile terminals have enhanced humans’ experience on spatial environment by improving their perception, understanding, and ability of control towards certain places. This paper centers on the advanced efforts of the augmentation technologies in intelligent sensing, cognition, and feedback, by introducing several innovative applications, including  1) the live-audio augmented project for the Washington Monument; 2) the CityScope project, developed by the MIT City Science Lab, which provides intelligent decision-making aid for non-professional users; and 3) Nature-X, an augmented reality application that creates nature community scenes based on Location Based Service. All these studied cases of spatial and environmental planning and design look into the future opportunities and challenges in the context of intelligent interaction.

Key words

Intelligent Interactive Landscape; Wearable Device; Experience Augmentation; Augmented Reality

第三類智能

A Third Intelligence

作者：耿百利，張子豪 Bradley CANTRELL, Zihao ZHANG

摘要

將人工智能應用于景觀設計學科所面臨的首要挑戰在于當前人工智能的定義不適用于系統性的景觀框架。一般的人工智能定義不關注復雜的生態關系，而是傾向于強調個體智能，忽視了人類和非人類智能體相結合而產生的智能。我們認為，在將人工智能充分應用于景觀設計領域之前，針對景觀設計學科制定一個切實的“智能”的定義非常必要。我們采用智能體－環境框架來定義景觀中的智能，并認為這一定義必須具體而明晰：在探討智能時，明確智能體、環境和首要目標是十分必要的。從智能的角度來看，設計師通過精心設計分布在環境中的各種智能體來創造景觀。而在將人工智能引入景觀設計學科的過程中，我們提出了與人類及非人類智能體共同演進的“第三類智能”。而這些智能形式間的相互作用與對話也將為景觀設計學科帶來新的機遇。

關鍵詞

人工智能；景觀設計學；智能體－環境框架；廣義智能；第三類智能

Abstract 

The fundamental challenge in the application of Artificial Intelligence to the discipline of Landscape Architecture is that current definitions of Artificial Intelligence do not fit within systemic landscape frameworks. Rather than focusing on complex ecological relationships, general definitions of Artificial Intelligence tend to emphasize the intelligence of individual entities and overlook the emergent intelligence of assemblages of human and non-human agents. We argue that it is important to develop a working definition of “intelligence” specific to Landscape Architecture before seriously considering the fruitful use of Artificial Intelligence in the production of environments. We adopt the agent-environment framework for defining intelligence in the context of landscape and assert that the definition has to be specific and situated: when discussing intelligence, it is necessary to clarify the agents, the environments, and the overarching goals. Taking intelligence as a lens, designers choreograph the intelligence distributed among human and non-human agents in the environment to produce landscapes. Introducing Artificial Intelligence to Landscape Architecture proposes a “third intelligence,” co-evolving with human and non-human actors. For Landscape Architecture, opportunity lies in the interactions and dialogues between these forms of intelligence.

Key words

Artificial Intelligence; Landscape Architecture; Agent-Environment Framework; Universal Intelligence; Third Intelligence

人工智能對設計的影響

The Impacts of Artificial Intelligence on Design

作者：劉瑜 Yu LIU

摘要

本次訪談圍繞人工智能及其可能的設計應用展開。受訪人劉瑜深入淺出地闡述了數據在人工智能領域的作用以及深度學習的邊界問題，并就人工智能在現階段能夠發揮的輔助設計功能及其未來發展趨勢進行了探討。劉瑜認為，通過深度學習，人工智能在相對封閉的領域內表現出色，但相比人類充滿創造力和倫理性的開放式思維，它仍存在局限，特別是在介于感性和理性之間的設計領域，它無法取代人類設計師進行設計。盡管如此，設計師可以把一些常規的基礎性工作交由人工智能去完成，從而在設計創新和用戶溝通上投入更多精力。

關鍵詞

人工智能；數據；深度學習；設計；黑箱；白箱

Abstract 

This interview centers on Artificial Intelligence and its possible applications in design fields. Yu Liu, the interviewee, explains the role of data in Artificial Intelligence, the boundary for Deep Learning, Artificial Intelligence’s function to aid design, and its future development. Liu also discusses how Artificial Intelligence can work within some particular fields, and the creative and ethical limits of Artificial Intelligence, especially in the design fields, which are between sensibility and rationality. Besides, he explains why Artificial Intelligence cannot take the place of human designers. In spite of this, Artificial Intelligence can be used to do repetitive or routine tasks so that designers can put more focus on design innovation and user experience optimization.

Key words

Artificial Intelligence; Data; Deep Learning; Design; Black Box; White Box

設計視角下人工智能的定義、應用及影響

Definition, Application and Influence of Artificial Intelligence on Design Industries

作者：蔡凌豪，范凌，賴文波，龍瀛，王鵬，辛向陽， Linghao CAI, Ling FAN, Wenbo LAI, Ying LONG, Peng WANG, Xiangyang XIN 

摘要

“人工智能”的出現大幅提升了人們的生產及生活效率，但與此同時，人類自身的就業環境也深受影響。那么，對于設計行業而言，人工智能又將帶來怎樣的挑戰與機遇？由此，《景觀設計學》邀請了6位來自建筑設計、城市規劃、景觀設計、工業設計等不同學科的學者、設計師，分別回答了什么是人工智能、人工智能可能對設計師的工作產生何種影響，以及人工智能會創造什么樣的生活方式三個問題。多數受訪者認為，當前的人工智能并非真正意義上的人工智能，其不具備自我意識，亦無法完成創造性行為。而在參與設計工作時，人工智能雖然可以大幅減少設計師的程序性勞動，但由于其采用的是權重疊加和“去少存多”的數據處理方式，運算結果缺乏倫理性和價值評判，因而還遠遠無法勝任創造性工作。而在未來，人工智能無疑會對人類的生活方式產生巨大影響，甚至遠超出我們的想象。

關鍵詞

人工智能；設計；機器；數據

Abstract 

Artificial Intelligence significantly promotes humans’ production efficiency and facilitates our daily life. Meanwhile, the climate of people’s employment is also under great impact. Through a group interview with six scholars and designers from the fields of Architecture, Urban Planning, Landscape Architecture, and Industrial Design, Landscape Architecture Frontiers attempts to provoke public attention on the challenges and opportunities (would be) brought by Artificial Intelligence by asking three questions: What is Artificial Intelligence? How would it influence designers’ working process and final results? And, what lifestyles would we have in the future under the influence of Artificial Intelligence? Most interviewees agree that, though Artificial Intelligence has largely helped us lessen workload on repetitive or routine tasks, nowadays it can neither be intelligent enough to have self-consciousness or perform creative jobs, nor offer ethical solutions or value judgments, because it is operated under weighted computing which is programmed by the majority rule. However, all the interviewees believe that Artificial Intelligence will make a big change on people’s future lifestyles, far beyond our imagination.

Key words

Artificial Intelligence; Design; Machine; Data

機器人自動化建造的景觀—自然、運算，以及自動化地形建模中的設計空間

Robotic Landscapes— Nature, Computation, and the Design Space of Autonomous Terrain Modeling

作者：克里斯托弗·吉魯特，伊爾瑪·赫爾克斯肯斯 Christophe GIROT, Ilmar HURKXKENS

摘要

2017年秋，蘇黎世聯邦理工學院與瑞士國家科研能力中心數字化制造小組就瑞士提契諾州坎頓地區的一條高速公路進行了實驗性設計。該項目展示了一系列運用計算機程序和機器人原理研發的設計，基于覆蓋整個提契諾河谷的激光雷達點云數據集的景觀模型為項目中所有的地形塑造提供了基礎。經過歷時15周的探索后，項目獲得了可喜的成果，同時展現出了一種通過機器人自動化設計來構思景觀的新途徑。

關鍵詞

機器人自動化建造的景觀；自動化地形建模；點云數據集；人類世

Abstract 

An experimental studio on a highway site in Canton of Ticino in Switzerland held at the ETH in the fall of 2017 is the result of a collaborative project with the National Center of Competence in Research Digital Fabrication, ETH Zürich. The work shows a series of designs that were developed through procedural and robotic principles. The landscape models based on a Lidar point cloud data set of the entire Ticino Valley served as the basis of all terrain operations. The results obtained after a 15-week studio are encouraging and show the way towards a new way of conceiving landscapes through robotic design.

Key words

Robotic Landscape; Autonomous Terrain Modeling; Point Cloud Data Set; Anthropocene

基于全球性城市中地方社區的數據體驗

The Data Experience of Local Neighborhoods in Global Cities 

作者：克里斯蒂安·德里克斯，露西·赫爾墨，法比奧·加里西亞，亞歷山大·卡奇凱威 Christian DERIX, Lucy HELME, Fabio GALICIA, Alexander KACHKAEV

摘要

本文著重介紹了一款用于評估城市環境狀況的搜索引擎平臺—CIVITAS。該平臺由伍茲貝格建筑設計事務所的SUPERSPACE團隊設計和開發，可幫助用戶判斷某處地域的宜居性或城市體驗是否符合其預期或特殊要求。基于空間特征、空間連通性和土地利用密度三類指標，用戶可從大量的社會空間數據集中選取所需數據，對整個城市或社區、建筑物甚至具體樓層進行定制化分析。CIVITAS適用于不同的城市，其內部匯編了一套服務于各項城市指標的數據庫，并正在將越來越多的全球性城市納入其中。本文呈現了利用此平臺在全球性城市的地方社區中所開展的研究。

關鍵詞

數據；CIVITAS搜索引擎；宜居性；可視化；指標

Abstract 

This article introduces CIVITAS, a search engine for urban conditions designed and developed by SUPERSPACE of Woods Bagot to allow stakeholders to identify qualities of liveability and urban experiences that suit their tacit desires and explicit requirements. Large data sets of socio-spatial quantities are selectable to create bespoke analytics across scales from whole city to neighborhoods, buildings, and even floors based on three categories: spatial character, connectivity, and land-use densities. CIVITAS is applicable across cities and a database for urban metrics has been compiled in-house that contains an increasing number of global cities. This article showcases research into neighborhoods within global cities using the platform.

Key words

Data; CIVITAS Search Engine; Liveability; Visualization; Metrics

StreeTalk慢行導航系統

Streetalk: A Navigation System for Pedestrians and Cyclists

作者：劉瀏，張帆，周博磊，王舟童，李穎欣 Liu LIU, Fan ZHANG, Bolei ZHOU, Zhoutong WANG, Yingxin LI

摘要

不同于當前以出行效率為主要考量因素的傳統導航系統，StreeTalk慢行導航系統更注重行人和騎行者的感受，特別是對安全感和舒適度的需求。通過運用物體檢測識別及場景語義分割等技術，城市街景的深度圖像特征被提取出來。之后，結合深度學習模型，機器可以預判人類對周邊環境的感知，并為數以千萬計的街景照片評分，從而建立起可以直觀顯示街道安全和舒適程度的導航系統。相關技術不僅可以提供定制化、全方位的慢行導航服務，為城市研究及決策、城市景觀設計提供相關依據，也為未來城市生活創造了更多可能。

關鍵詞

城市街道；安全感與舒適度；慢行導航系統；深度學習；物體檢測識別；場景語義分割

Abstract 

Different from the conventional efficiency-driven navigation systems, StreeTalk navigation system is developed for pedestrians and cyclists to optimize their travelling safety and comfort. Through the application of technologies including object detection and scene parsing, the characteristics of urban streetscapes can be extracted. By combining deep learning models, machines can imitate human’s perception and evaluate streetscapes automatically, and intuitively show users the street safety and comfort level of different commuting options. Relevant technologies can not only be applied in offering panorama navigation services for pedestrians and cyclists, but also better support urban research, inform decision-making and urban landscape design, and explore more possibilities for future urban life. 

Key words

Urban Street; Safety and Comfort; Navigation System for Pedestrians and Cyclists; Deep Learning; Object Detection; Scene Parsing

基于空間句法模型的數據化城市設計—以吉林市朝陽廣場設計為例

The Application of Space Syntax Modeling in Data-Based Urban Design— An Example of Chaoyang Square Renewal in Jilin City

作者：盛強，周晨，凱萬·卡里米，路安華，邵敏 Qiang SHENG, Chen ZHOU, Kayvan KARIMI, Anhua LU, Min SHAO

摘要

空間句法是研究城市空間與社會經濟活動之間關系的一系列理論與技術，大數據時代便捷的數據獲取使得空間句法研究成果能夠快速應用于數據分析與設計實踐。本文以吉林省吉林市軌道交通站點之一的朝陽廣場為例，通過應用線段模型以及視域模型中的多智能體模擬工具，對廣場內部及周邊街道的交通流量和視域整合度進行了分析，并呈現了將分析結果應用于廣場設計及方案評價的全過程，以期提升廣場空間的使用率；本文同時展望了機器學習和人工智能等新技術在推進空間句法基礎研究和設計應用方面的前景。

關鍵詞

空間句法；數據化設計；多源數據；交通流量；視域整合度；多智能體

Abstract 

In the past decades, Space Syntax offers a series of theories and techniques to study the relationship between urban space and social-economic activities, and has been proved effective in analysis and design practices thanks to the open sources in the big data era. Taking the Chaoyang Square Renewal project in Jilin City, Jilin Province as an example, this article introduces the analyses of traffic volumes and visual integration of the square and the connected streets with modeling tools such as Segment Map and the intelligent multi-agent systems in Visibility Graph Analysis. All these analyses provided a basis for the full design process, from conceptual design to proposal evaluation, in order to activate this site through introducing pedestrian vitality. Prospects on new technologies in Artificial Intelligence, such as machine learning, are also explored to promote the research of Space Syntax and related application in urban design.

Key words

Space Syntax; Data-Based Design; Multi-Source Data; Traffic Flow; Visual Integration; Intelligent Multi-Agent System

“多些答復，少些噱頭”—致“步道實驗室”主持的多倫多更新項目

“More Buzzwords than Answers”— To Sidewalk Labs in Toronto

作者：瑪麗亞娜·瓦爾韋德，亞歷山大·弗林 Mariana VALVERDE, Alexandra FLYNN

摘要

近來，各大主流媒體和技術相關領域爭相報道了美國基礎設施公司“步道實驗室”將在加拿大首都多倫多建造一座新型高科技社區（即一項名為“步道多倫多”的項目）。總的來說，國際評論界認為，在公眾對建造“智慧城市”項目看法不一的情況下，政府將城市規劃的諸項決策權下放至一家以數據為主營內容的私營企業的做法，可謂是機遇與風險并存。

與大多數文章籠統談及智慧城市和技術導向型項目的視角不同，本文旨在以“步道多倫多”項目為具體案例，提出下列5個問題：其一，作為私企的步道實驗室（其為互聯網巨鱷“谷歌”的姊妹公司）與推進城市水岸振興的政府公共機構“濱水多倫多”達成合作協議的過程缺乏透明度，且作為推動項目進行的公共合作方，濱水多倫多也未能向市民作出負責任的解釋說明；其二，步道實驗室對于其將在哪些地塊開展更新項目的表述含混不清，使得公眾對該私營技術公司并未足夠重視多倫多當地規劃法規和房地產現狀的情況表示擔憂；其三，作為一項由數據技術公司積極推進的城市更新規劃項目，人們不免懷疑掌控數據信息才是步道實驗室的真正意圖；其四，這一案例呈現了城市更新項目中合同簽訂與開發權給予機制的相關問題；最后，正如文中所述，盡管多倫多市議會仍未獲知步道實驗室與濱水多倫多雙方所簽訂協議的內容，但已經公布的該項目至今為止的推進方式，以及協議簽訂雙方發表的聲明，已足以引發公眾對于企業和政府責任感的強烈質疑。

關鍵詞

公私合營；專項機構；透明度；擔責

Abstract 

Many articles have appeared in mainstream media and in tech-oriented venues about Sidewalk Labs’ ideas for a new high-tech neighbourhood in Toronto (a project named Sidewalk Toronto). By and large, international commentary has focused on the opportunities and risks of giving over control over many city planning decisions to a private data-oriented corporation, with people lining up for or against “smart city” ideas, in general.

This article will set aside generalities about “smart cities” and technology, and instead pose a few questions about the particulars of Sidewalk Toronto project. The first question concerns the striking lack of transparency of the agreement between Sidewalk Labs (a Google sister company) and Waterfront Toronto, the public authority promoting the project, which is not directly accountable to the city or the citizens. The second question concerns the equally striking ambiguity about which parcel of land is being sought by Sidewalk Labs — an ambiguity that suggests a worrying lack of concern, on the tech company’s part, about both local planning law and local real estate realities. The third set of concerns is about the ownership of the data that appears to be Sidewalk Labs’ real interest. Fourthly, problems in the contract award and procurement mechanisms will be raised. Finally, even though the agreement has not yet been seen even by city council, the process so far and the statements by both parties raise serious concerns about accountability, the fifth point raised in this article.

Key words

Public-Private Partnerships; Special Purpose Agencies; Transparency; Accountability

設計未來的理想街道

Let Us Design Streets for the Future We Want

作者：安基塔·查克拉，梅琳達·漢森 Ankita CHACHRA, Melinda HANSON

摘要

本文旨在呼吁人們關注街道設計在引入和優先考慮某些用途時的作用，并突顯了世界各個城市為將街道從汽車導向轉變為居民導向所付出的努力。身處當今時代，我們對于未來街道的暢想被無人駕駛汽車的問世所蒙蔽和限制，卻忽視了利用低技術的解決方案有效改善城市環境的可能。本文呼吁人們盡快采取行動以改變當下盛行的街道設計方法，并將衡量項目成功與否的指標由以往的汽車導向型指標轉換為強調可達性、安全性、空間公平分配、環境質量、公共健康和整體生活質量的指標。從業者和決策者不應想當然地認為當今城市面臨的所有挑戰都可以通過技術來解決，而應著重發展那些采納了優選高效通勤方式且有助于創建市民友好型城市的街道設計。

關鍵詞

街道設計；公平城市；社區參與；公共生活；道路安全

Abstract 

This article will call attention to the role of street design in inviting and prioritizing certain uses and highlight efforts made by cities across the globe to move away from car-oriented and toward people-oriented streets. There are many low-technology solutions available to create better cities, and yet, we are in an era where the vision for the future of streets is clouded by the advent of autonomous vehicles. This article will emphasize that urgent action is needed to change the prevailing approach to street design and shift the measure of success away from car-oriented metrics and toward metrics that address access, safety, equitable distribution of space, environmental quality, public health, and overall quality of life. Practitioners and decision-makers should not assume that technology will solve the challenges cities are facing today and should focus first on designing streets that prioritize the most efficient modes of transport and create people-friendly cities.

Key words

Street Design; Equitable Cities; Community Engagement; Public Life; Road Safety