Self-Driving Universities: An Autonomous, Formally Verified Multi-Agent Framework for Dynamic Curriculum and Resource Optimization

University timetabling and resource allocation automation has historically relied on rigid operations research models that scale poorly during sudden operational disruptions. While recent advances in multi-agent systems (MASs) and large language models (LLMs) offer dynamic semantic adaptability, these purely probabilistic architectures fail to satisfy strict, NP-hard combinatorial constraints, rendering them unsafe for independent administrative execution. This research introduces a hybrid agentic-symbolic framework that bridges this evaluation gap by decoupling neural reasoning from deterministic execution. The architecture utilizes a network of specialized LLM agents to negotiate institutional demand. Meanwhile, a Z3 formal verification engine acts as an immutable gatekeeper, mathematically blocking any physically impossible schedule from executing. To evaluate hardware feasibility, the framework underwent an 8-semester longitudinal simulation across two distinct institutional topologies (totaling 16 simulated semesters) across heterogeneous computing environments, incorporating stochastic structural and personnel shocks. Empirical telemetry indicates a critical failure mode within constrained 23 Edge CPU infrastructure, which we characterize as autoregressive token explosion, in which 24 generative model’s attention mechanism collapses during the production of large schemas, paralyzing the system. To address this, the architecture substitutes conventional substantial solvers with a sub-millisecond greedy heuristic fallback. The results demonstrate that this deterministic safety mechanism intercepts catastrophic neural failures and resolves the scheduling matrix in under 1.0 millisecond. Ultimately, this study demonstrates that formally verified multi-agent systems can attain zero-violation compliance and uninterrupted institutional continuity when supported by lightweight, hardware-aware heuristic overrides.

Authors

Institutions

Publication Details

Journal
Applied Sciences
Published
2026-11-18
DOI
https://doi.org/10.3390/app162211408
Primary Topic
Multi-Agent Systems and Negotiation
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Self-Driving Universities: An Autonomous, Formally Verified Multi-Agent Framework for Dynamic Curriculum and Resource Optimization

Ashraf M Ali
Applied Sciences
Multi-Agent Systems and Negotiation
article

Self-Driving Universities: An Autonomous, Formally Verified Multi-Agent Framework for Dynamic Curriculum and Resource Optimization

Ashraf M Ali
article en

Abstract

University timetabling and resource allocation automation has historically relied on rigid operations research models that scale poorly during sudden operational disruptions. While recent advances in multi-agent systems (MASs) and large language models (LLMs) offer dynamic semantic adaptability, these purely probabilistic architectures fail to satisfy strict, NP-hard combinatorial constraints, rendering them unsafe for independent administrative execution. This research introduces a hybrid agentic-symbolic framework that bridges this evaluation gap by decoupling neural reasoning from deterministic execution. The architecture utilizes a network of specialized LLM agents to negotiate institutional demand. Meanwhile, a Z3 formal verification engine acts as an immutable gatekeeper, mathematically blocking any physically impossible schedule from executing. To evaluate hardware feasibility, the framework underwent an 8-semester longitudinal simulation across two distinct institutional topologies (totaling 16 simulated semesters) across heterogeneous computing environments, incorporating stochastic structural and personnel shocks. Empirical telemetry indicates a critical failure mode within constrained 23 Edge CPU infrastructure, which we characterize as autoregressive token explosion, in which 24 generative model’s attention mechanism collapses during the production of large schemas, paralyzing the system. To address this, the architecture substitutes conventional substantial solvers with a sub-millisecond greedy heuristic fallback. The results demonstrate that this deterministic safety mechanism intercepts catastrophic neural failures and resolves the scheduling matrix in under 1.0 millisecond. Ultimately, this study demonstrates that formally verified multi-agent systems can attain zero-violation compliance and uninterrupted institutional continuity when supported by lightweight, hardware-aware heuristic overrides.

Applied SciencesVol. 16(22)
Arab Open University (BH)
Multi-Agent Systems and Negotiation
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon