AI for Happiness: Pedestrian Path Generation through Agent-Based Simulations with Deep Reinforcement Learning

Ayse Glass; Burak Bek; Eylul K. Menges; Iryna Okhrin; Jörg Noennig; Mariela Rossana Sanchez; Pramod Baddam; René Jäkel; Roman Glass

AI for Happiness: Pedestrian Path Generation through Agent-Based Simulations with Deep Reinforcement Learning

Ayse Glass, Burak Bek, Eylul K. Menges, Iryna Okhrin, Jörg Noennig, Mariela Rossana Sanchez, Pramod Baddam, René Jäkel, Roman Glass

Abstract

AI integration in Smart Cities, primarily through agent-based simulations,
holds transformative potential for understanding and enhancing citizen
behavior. Striking a balance between complexity and computational
feasibility is essential. Our research question is, how can we make agents
behave more realistically? We assumed that happiness is a motivating factor
for the mobility. Insights from a survey of 130 citizens inform our weightings.
We used reinforcement learning (RL) as a method and Q-learning as an
algorithm to generate a baseline, further enhanced with neural networks for
adaptability. This study contributes to data-driven urban design by offering
efficient intelligent agent solutions. The research lays foundations for smart
agents in urban design, which can be used to generate synthetic data.

Original language	English
Pages	139
Number of pages	1
Publication status	Published - 10 Oct 2023
Event	Digital Total: Computing & Data Science an der Universität Hamburg und in der Wissenschaftsmetropole Hamburg - Universität Hamburg, Hamburg, Germany Duration: 9 Oct 2023 → 10 Oct 2023

Conference

Conference	Digital Total
Country/Territory	Germany
City	Hamburg
Period	9/10/23 → 10/10/23

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 11 Sustainable Cities and Communities

Keywords

Artificial intelligence
Computation
Digital Tools
pedestrian orientation
pedestrian route networks
Urban analysis
data science

Cite this

@conference{7f64ac157eee4012a86c6a735072cd52,

title = "AI for Happiness: Pedestrian Path Generation through Agent-Based Simulations with Deep Reinforcement Learning",

abstract = "AI integration in Smart Cities, primarily through agent-based simulations, holds transformative potential for understanding and enhancing citizen behavior. Striking a balance between complexity and computational feasibility is essential. Our research question is, how can we make agents behave more realistically? We assumed that happiness is a motivating factor for the mobility. Insights from a survey of 130 citizens inform our weightings. We used reinforcement learning (RL) as a method and Q-learning as an algorithm to generate a baseline, further enhanced with neural networks for adaptability. This study contributes to data-driven urban design by offering efficient intelligent agent solutions. The research lays foundations for smart agents in urban design, which can be used to generate synthetic data.",

keywords = "Artificial intelligence, Computation, Digital Tools, pedestrian orientation, pedestrian route networks, Urban analysis, data science",

author = "Ayse Glass and Burak Bek and Menges, \{Eylul K.\} and Iryna Okhrin and J{\"o}rg Noennig and Sanchez, \{Mariela Rossana\} and Pramod Baddam and Ren{\'e} J{\"a}kel and Roman Glass",

year = "2023",

month = oct,

day = "10",

language = "English",

pages = "139",

note = "Digital Total : Computing \& Data Science an der Universit{\"a}t Hamburg und in der Wissenschaftsmetropole Hamburg ; Conference date: 09-10-2023 Through 10-10-2023",

}

TY - CONF

T1 - AI for Happiness

T2 - Digital Total

AU - Glass, Ayse

AU - Bek, Burak

AU - Menges, Eylul K.

AU - Okhrin, Iryna

AU - Noennig, Jörg

AU - Sanchez, Mariela Rossana

AU - Baddam, Pramod

AU - Jäkel, René

AU - Glass, Roman

PY - 2023/10/10

Y1 - 2023/10/10

N2 - AI integration in Smart Cities, primarily through agent-based simulations, holds transformative potential for understanding and enhancing citizen behavior. Striking a balance between complexity and computational feasibility is essential. Our research question is, how can we make agents behave more realistically? We assumed that happiness is a motivating factor for the mobility. Insights from a survey of 130 citizens inform our weightings. We used reinforcement learning (RL) as a method and Q-learning as an algorithm to generate a baseline, further enhanced with neural networks for adaptability. This study contributes to data-driven urban design by offering efficient intelligent agent solutions. The research lays foundations for smart agents in urban design, which can be used to generate synthetic data.

AB - AI integration in Smart Cities, primarily through agent-based simulations, holds transformative potential for understanding and enhancing citizen behavior. Striking a balance between complexity and computational feasibility is essential. Our research question is, how can we make agents behave more realistically? We assumed that happiness is a motivating factor for the mobility. Insights from a survey of 130 citizens inform our weightings. We used reinforcement learning (RL) as a method and Q-learning as an algorithm to generate a baseline, further enhanced with neural networks for adaptability. This study contributes to data-driven urban design by offering efficient intelligent agent solutions. The research lays foundations for smart agents in urban design, which can be used to generate synthetic data.

KW - Artificial intelligence

KW - Computation

KW - Digital Tools

KW - pedestrian orientation

KW - pedestrian route networks

KW - Urban analysis

KW - data science

UR - https://www.hcds.uni-hamburg.de/en/current/all-events/digital-total/digital-total-boa.pdf

M3 - Meeting Abstract

SP - 139

Y2 - 9 October 2023 through 10 October 2023

ER -

AI for Happiness: Pedestrian Path Generation through Agent-Based Simulations with Deep Reinforcement Learning

Abstract

Conference

UN SDGs

Keywords

Other files and links

Cite this