kth.sePublications
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Multi-Agent Control in Warehousing: A Deep Q-Network Approach
KTH, School of Electrical Engineering and Computer Science (EECS).
KTH, School of Electrical Engineering and Computer Science (EECS).
2023 (English)Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
Abstract [en]

With an increase in consumption, warehouses increase in size and demand for fastdistribution of goods. One solution to this problem is self learning robots that can adapt toany warehouse. This project implements the Deep Reinforcement Learning (DRL) algorithmDeep Q-Network (DQN) to a simulated warehouse in order to achieve close to optimalperformance. In order to verify the functionality of the written DQN algorithm, animplementation on a small single agent environment was conducted. After sufficient learningcapabilities were proven, a multi-agent system was developed. During training the multi-agent implementation showed promising returns and few collisions with other agents andwalls. When testing the trained systems, agents trained in small environments receivedlarger returns than those trained in a larger environment. Although the developed DQNstructure indicated learning capabilities, improvements can be made.

Abstract [sv]

Med ökande konsumtion ökar även storleken av lagerlokaler och kravet på en ökaddistribution av varor. En lösning till detta problem är självlärande robotar som kan anpassasig till vilken lagerlokal som helst. Det här projektet implementerar Deep ReinforcementLearning (DRL) algoritmen Deep Q-network (DQN) i en simulerad lagerlokal för att kunnauppnå nära inpå optimal prestanda. För att kunna verifiera att den skrivna DQN algoritmenfungerar som den ska, implementeras den på en liten miljö med en agent. Efter uppvisadtillräcklig inlärningsförmåga, utvecklades ett multi-agent system. Under inlärning visadesystemet lovande resultat och få krockar med andra agenter och väggar. När de tränadesystemen testades, gav de som tränats i en mindre miljö bättre resultat än de som tränats ien stor miljö. Även om DQN algoritmen visade inlärningsförmågor, finns potentiellaförbättringar.

Place, publisher, year, edition, pages
2023. , p. 123-129
Series
TRITA-EECS-EX ; 2023:144
National Category
Electrical Engineering, Electronic Engineering, Information Engineering
Identifiers
URN: urn:nbn:se:kth:diva-341487OAI: oai:DiVA.org:kth-341487DiVA, id: diva2:1821887
Supervisors
Examiners
Projects
Kandidatexjobb i elektroteknik 2023, KTH, StockholmAvailable from: 2023-12-21 Created: 2023-12-21

Open Access in DiVA

fulltext(211487 kB)33 downloads
File information
File name FULLTEXT01.pdfFile size 211487 kBChecksum SHA-512
69786101c351a58f7bd524c3aeee40c661028b577366c4a725033372b88c624c87c2183b6acca2d3d43bbd2bb2f3942326c69263e70c99cf1db027ce9c4e9ae2
Type fulltextMimetype application/pdf

By organisation
School of Electrical Engineering and Computer Science (EECS)
Electrical Engineering, Electronic Engineering, Information Engineering

Search outside of DiVA

GoogleGoogle Scholar
Total: 33 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 136 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf