Multi-Agent DSL Framework for Intelligent Task Scheduling

Quick Links

• Live Demo: https://max-yuan-22.github.io/Agent_DSL/
• Paper (PDF): overleaf_ready/finalpaper.pdf
• Code: https://github.com/Max-YUAN-22/Agent_DSL
• Data: data/
• Repro Steps: REPRODUCIBILITY.md

📚 Overview

This repository presents a Multi-Agent Domain-Specific Language (DSL) framework for optimizing task scheduling and resource allocation across various domains. The framework integrates three core algorithms:

• ATSLP: Adaptive Task Scheduling with Load Prediction
• HCMPL: Hierarchical Coordination with Multi-Path Learning
• CALK: Collaborative Agent Learning and Knowledge Sharing

🎯 Core Algorithms

• ATSLP: Adaptive Task Scheduling with Load Prediction
• HCMPL: Hierarchical Coordination with Multi-Path Learning
• CALK: Collaborative Agent Learning with Knowledge Sharing

📊 Live Demo

🌐 View Live Demo

Demo Scenarios:

• 🚨 [Emergency Response]
• 🔥 [Wildfire Response]
• ⚡ [Smart Grid Outage]
• 🏥 [Smart Healthcare]
• 🚦 [Traffic Management]
• 🤖 [Warehouse Robotics]
• 🛰️ [Urban Search & Rescue]

📈 Performance Results

• Throughput: 2.45 tasks/second
• Response Time: 783ms average
• Success Rate: 100% task completion
• Improvement: 43.9% throughput increase over baseline frameworks

📚 Academic Paper

The full research paper is available in overleaf_ready/finalpaper.tex.

🔬 Reproducibility

All experimental data and results are stored in the data/ directory:

• comprehensive_experimental_data.json
• real_api_benchmark_results.json
• honest_api_benchmark_results.json
• real_cache_performance.json

🚀 Smart City Application

A real-world smart city multi-agent system based on the framework:

Key Features:

• Master Agent: Core decision-making unit for city management
• Sub Agents: Manage tasks like Traffic 🚦, Weather 🌦️, Parking 🅿️, and Security 🔍
• Real-Time Sensor Data: API integration with actual sensors or simulations
• Task Tracker: Full lifecycle management and reporting

Running Locally:

# Start the local server
python3 -m http.server 8080

# Access the system
open http://localhost:8080

🛠️ Requirements

Refer to requirements.txt for Python dependencies.

📜 License

MIT License - see LICENSE for details.

🙏 Acknowledgments

We would like to express our sincere gratitude to Professor Hailong Shi from the Institute of Microelectronics, Chinese Academy of Sciences, for his valuable guidance and suggestions on project conception and technical roadmap. 感谢石海龙教授（中科院微电子所）在项目构思和技术路线方面提供的宝贵指导和建议。

---

For questions or more details, please refer to the academic paper or demo documentation.