
Pedro Luís Alves VelosoUniversity of Arkansas | U of A
Pedro Luís Alves Veloso
PhD in Computational Design / MSc in Architecture and Urbanism / Bachelor in Architecture and Urbanism
About
39
Publications
20,409
Reads
How we measure 'reads'
A 'read' is counted each time someone views a publication summary (such as the title, abstract, and list of authors), clicks on a figure, or views or downloads the full-text. Learn more
140
Citations
Introduction
Computational designer and architect. I am interested in the intersection of design with Cybernetics, classical AI, bio-inspired AI, agent-based modeling Reinforcement Learning and Machine learning. I design generative systems for design exploration and, currently, I investigate the application of AI in architectural composition and space planning.
Additional affiliations
August 2015 - present
August 2015 - present
August 2011 - December 2011
Education
January 2015 - May 2022
March 2008 - December 2010
March 2001 - December 2006
Publications
Publications (39)
This paper explores the optimization of apartment building layouts using five different meta-heuristic optimization algorithms: Simple Genetic Algorithm (SGA), Simple Evolutionary Algorithm (SEA), Self-Adaptive Differential Evolution (SADE), Particle Swarm Optimization (PSO), and Artificial Bee Colony (ABC). The study aims to compare and analyze th...
Motivated by reflection-in-action in architectural design, this article introduces a spatial synthesis artifact that relies on multi-agent reinforcement learning to address spatial goals with fine-grained control in a simulation. It relies on parameter sharing with proximal policy optimization and a parameterized reward function to train robust age...
Over the past years, computational methods based on deep learning—that is, machine learning with multilayered neural networks—have become state-of-the-art in main research areas in computer-aided architectural design (CAAD). To understand current trends of CAAD with deep learning, to situate them in a broader historical context, and to identify fut...
In this paper, we address the generation of realistic apartment complex layouts. We define this layout problem as finding the optimal locations and angles of the buildings within the site boundary according to the building codes, zoning regulations, and design preferences. In contrast to manual design heuristics, which are laborious and typically p...
Motivated by situated cognition in architectural design, this research investigates a computational approach for spatial synthesis that supports designers inside the generative loop with fine-grained interaction. After reviewing and creating a taxonomy of classic paradigms of generative modeling, we present a research guideline for interactive spat...
This research proposes an alternative representation of apartment building mass for developing an automation tool of apartment complex
layout design. An apartment building mass is a collection of residential units with required facilities and a constituent of an apartment
complex. The shape of mass changes according to the configuration of the unit...
This paper reports a pedagogical experience that incorporates deep learning to design in the context of a recently created course at the Carnegie Mellon University School of Architecture. It analyses an exercise called Bubble2Floor (B2F), where students design
floor plans for a multi-story row-house complex. The pipeline for B2F includes a paramet...
In this paper we propose a deep learning (DL) method to investigate existing types of high-rise buildings and to generate new ones. We collected data comprehending diverse forms of high-rise building from major cities in the world to train a generative DL model (IntroVAE) to capture morphological features. After clustering the features, we can dist...
In this paper we propose a deep learning (DL) method to investigate existing types of high-rise buildings and to generate new ones. We collected data comprehending diverse forms of high-rise building from major cities in the world to train a generative DL model (IntroVAE) to capture morphological features. After clustering the features, we can dist...
In this article we will focus on a branch of the work of CRAIDL that addresses generative design in the context of architectural configuration. Generative design is an indirect method of design where the designer employs models that embed some form of decision-making, such as instructions or behaviors, to generate design alternatives. With the curr...
This paper describes the Urban Structure Synthesizer (USS), a research prototype based on deep learning that generates diagrams of morphologically consistent urban fabrics from context-rich urban datasets. This work is part of a larger research on computational analysis of the relationship between urban context and morphology. USS relies on a data...
Over the last decades, a vast repertoire of computational methods has been employed for the synthesis of spatial configuration. Many of these techniques, such as the long-standing black box optimization or the recent generative adversarial networks, enable a quick exploration of the design solutions based on destructive operations, but encapsulate...
Agent-based models rely on decentralized decision making instantiated in the interactions between agents and the environment. In the context of generative design, agent-based models can enable decentralized geometric modelling, provide partial information about the generative process, and enable fine-grained interaction. However, the existing agent...
We present a novel method for generating building geometry using deep learning techniques based on contextual geometry in urban context and explore its potential to support building massing. For contextual geometry, we opted to investigate the building footprint, a main interface between urban and architectural forms. For training, we collected GIS...
Generative systems have a significant potential to synthesize innovative design alternatives. Still, most of the common systems that have been adopted in design require the designer to explicitly define the specifications of the procedures and in some cases the design space. In contrast, a generative system could potentially learn both aspects thro...
Under the umbrella of the digital turn, novel computational workflows and distinct aesthetical principles are becoming an integral part of architectural education. Nonetheless, in current educational settings, there is not much scope for a deep understanding nor the development of custom computational design methods beyond standard toolkits. To fil...
In this paper we review the research on multi-agent space planning (MASP) during the period of 2008-2017. By MASP, we refer to space planning (SP) methods based on online mobile agents that map local perceptions to actions in the environment, generating spatial representation. We group two precedents and sixteen recent MASP prototypes into three ca...
Generative systems have a significant potential to synthesize innovative design alternatives. Still, most of the common systems that have been adopted in design require the designer to explicitly define the specifications of the procedures and in some cases the design space. In contrast, a generative system could potentially learn both aspects thro...
This paper describes a design customization system that integrates two aspects of Computer-Aided Architectural Design (CAAD) that are usually developed in separate workflows: the algorithmic generation of designs and the detailed representation of the building. The system's workflow starts with the definition of shape grammar rules by an architect....
Parametric design is a computational process that is increasingly widespread in contemporary architecture and requires a technical domain that permeates the need to think of teaching methods and their application in design practices. Based on this observation, it is a possibility to deepen the debate about the potentials and limitations of the comp...
This paper investigates the relationship between computational and architectural design concepts since the late 1980s, with the aim of identifying whether it was effectively grounded on a real understanding of technology. We took The Art of Computer Graphics Programming, by Mitchell et al. (1987) as the starting point of our temporal scope. Next, w...
The use of adjacency graphs to represent and generate architectural arrangements tends to favor direct connections between contiguous rooms. These disregard specialized circulatory systems (such as corridors), which consider connections between non-contiguous spatial units or accesses. This paper addresses two specific issues: (1) how to represent...
This paper analyzes the evolution and trends of international conferences addressing Computer-Aided Architectural Design over the past five decades. It starts with the Design Methods Movement conferences, when the possible contributions of computers to Architectural Design started being discussed. It shows how new interests along the history of CAA...
Nós da Arquitetura, assim denominamos nossa contribuição
às questões que permeiam essa disciplina, que trata
da produção e interpretação do espaço. Em muitos níveis, o
termo ‘nós’ se mostrou apropriado para conceber o livro que
você tem agora em mãos.
Nós somos arquitetos, professores e pesquisadores, mas
nossos temas de interesse parecem, à primei...
This paper investigates the use of diagrams related to cybernetics and information theory in experimental design practices in the 1960s and 1970s. Those diagrams are investigated in light of Vilém Flusser's concept of game, which mediates the modus operandi of computers and possible strategies for design based on distributed cognition. The research...
In the 60's and 70's, Christopher Alexander faced important architectural dilemmas and his various solutions became relevant theoretical references to contemporary debates. Reviewing his design theory between the synthesis of form and the pattern languages , this paper addresses the challenge of Alexander to establish a systemic conception of space...
This article investigates the notion of an augmented architectural design culture elapsed of the digital revolution. Based on analytical categories proposed by the philosopher Vilém Flusser it compares two systems of creative production (the tool-architect and the robot-architect), establishing new perspectives on the architectural design supported...
This article deals with the relationship between design process and its support instruments. In contrast to an idealistic view of architectural creation that minimizes the importance of the design action, it supports the hypothesis that the design instruments, with its possibilities and limits, stimulate specific modes of conception. To this end, i...
As principais inovações da Arquitetura contemporânea ocorrem em estreita relação com o desenvolvimento e difusão da tecnologia digital, estabelecendo não só novas categorias, como também modos específicos de se conceber Arquitetura. Desse modo, um caminho bastante pertinente para se compreender os desafios e limites da produção contemporânea é o es...