Velocity Obstacles-Induced Steering Behavior: Emotional Reciprocal Velocity Obstacles and Flocking-Induced Velocity Obstacles Models for Multi-Agent Navigation

Fachri, Moch (2025) Velocity Obstacles-Induced Steering Behavior: Emotional Reciprocal Velocity Obstacles and Flocking-Induced Velocity Obstacles Models for Multi-Agent Navigation. Doctoral thesis, Institut Teknologi Sepuluh Nopember.

Text 7022211027-Dissertation.pdf - Accepted Version Restricted to Repository staff only until 1 April 2028. Download (7MB) | Request a copy

Abstract

Autonomous agents in virtual environments, such as crowd simulations and digital animations, require precise steering behaviors to navigate complex spaces while avoiding collisions. This task becomes more challenging when agents must operate as a group, balancing individual movement with overall cohesion. Traditional methods address this by defining external steering behaviors, which are then adjusted using the Velocity Obstacles (VO) framework. However, this approach limits adaptability, particularly in real-time scenarios that demand fluid inter-agent coordination. This dissertation introduces a novel method where steering behaviors are directly generated within the VO process, enabling agents to dynamically adapt to group navigation needs without external inputs. By embedding behaviors such as leader-following and flocking into the VO framework, this approach enhances multi-agent coordination, improving responsiveness and reducing disruptions. In leader-following, agents adjust their spacing based on a stress-based safety mechanism that prioritizes the leader’s movement. For flocking, the Flocking-induced Velocity Obstacles (FIVO) method incorporates alignment, cohesion, and separation principles to promote unified movement through velocity consensus. Simulation results show that ERVO reduces interaction overhead by 0.753 seconds and collision rates by 18.81% in leader-following scenarios, while FIVO supports cohesive group movement. This approach represents a significant advancement in multi-agent navigation, improving fluidity and coordination in real-time group interactions within virtual environments.
==================================================================================================================================
Agen otonom dalam lingkungan virtual, seperti simulasi keramaian dan animasi digital, memerlukan steering behavior yang presisi untuk menavigasi ruang yang kompleks dan menghindari tabrakan. Tantangan ini semakin besar ketika agen beroperasi dalam grup, menyeimbangkan kebutuhan pergerakan individu dengan kohesi grup. Secara tradisional, masalah ini diatasi dengan mendefinisikan steering behavior eksternal yang disesuaikan melalui kerangka kerja Velocity Obstacles (VO). Namun, ketergantungan pada perilaku eksternal membatasi adaptabilitas, terutama dalam situasi yang memerlukan responsif waktu nyata dan koordinasi antar-agen yang dinamis. Disertasi ini memperkenalkan pendekatan inovatif dengan menghasilkan steering behavior langsung dalam proses VO, memungkinkan agen beradaptasi dinamis dengan tuntutan navigasi grup tanpa input eksternal. Pendekatan ini meningkatkan koordinasi multi-agen dan mengurangi gangguan dalam navigasi lingkungan virtual yang kompleks. Studi ini berfokus pada dua steering behavior: leader-following dan flocking. Dalam leader-following, seorang leader memandu follower menuju tujuan bersama, dengan Emotional Reciprocal Velocity Obstacles (ERVO) memastikan jarak optimal antar agen. Untuk flocking, Flocking-induced Velocity Obstacles (FIVO) menggabungkan aturan keselarasan, kohesi, dan pemisahan dalam VO, mendorong pergerakan terpadu. Hasil simulasi menunjukkan ERVO mengurangi interaction overhead sebesar 0,753 detik dan tabrakan sebesar 18,81% dalam leader-following, sementara FIVO mendukung pergerakan grup yang kohesif. Pendekatan berbasis VO ini menandai kemajuan signifikan dalam navigasi multi-agen, meningkatkan koordinasi dalam interaksi grup waktu nyata.

Item Type:	Thesis (Doctoral)
Uncontrolled Keywords:	Autonomous agents, Crowd Simulation, Emotional Reciprocal Velocity Obstacles, Flocking, Flocking-induced Velocity Obstacles, Group navigation, Leader-following, Multi-Agent Navigation, Reciprocal Velocity Obstacles, Steering behavior, Velocity obstacles.
Subjects:	Q Science > QA Mathematics > QA336 Artificial Intelligence Q Science > QA Mathematics > QA76.9 Computer algorithms. Virtual Reality. Computer simulation. T Technology > T Technology (General) > T57.62 Simulation
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20001-(S3) PhD Thesis
Depositing User:	Moch Fachri
Date Deposited:	31 Jan 2025 06:32
Last Modified:	18 Mar 2025 04:12
URI:	http://repository.its.ac.id/id/eprint/117296

Actions (login required)

View Item