Changelog
This document tracks all notable changes to OmniEmbodied.
The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.
[1.0.0] - 2025-08-07
Initial Release
This is the first public release of OmniEmbodied: A Comprehensive Benchmark for Embodied Multi-Agent Systems.
Added - Core OmniSimulator simulation engine with text-based environment representation - Comprehensive EAR-Bench evaluation suite with 1,500 scenarios - Multi-agent evaluation framework supporting single-agent and multi-agent tasks - LLM integration supporting multiple providers (OpenAI, DeepSeek, Qwen, etc.) - Task categories: tool reasoning, implicit collaboration, compound reasoning - Automated data generation pipeline for scenario creation - Configuration management system with YAML-based configuration - Documentation with Sphinx and Read the Docs integration - Evaluation metrics for success rates, step efficiency, and performance analysis - Examples and tutorials for quick start
Core Features - OmniSimulator Engine: Text-based simulation with physical constraints and dynamic capabilities - EAR-Bench Dataset: 1,500 diverse scenarios across household and industrial domains - Multi-Agent Support: Single-agent and multi-agent coordination scenarios - LLM Integration: Support for various language model providers with standardized interfaces - Evaluation Framework: Comprehensive benchmarking with detailed performance metrics - Data Generation Tools: Automated scenario generation and task creation pipeline
Task Categories - Tool reasoning: Dynamic capability acquisition and tool usage - Implicit collaboration: Autonomous coordination without explicit directives - Compound reasoning: Complex multi-step problem solving - Physical reasoning: Understanding constraints and spatial relationships
Technical Highlights - Text-based environment representation for LLM compatibility - Dynamic capability system allowing agents to acquire tools as needed - Comprehensive action validation and execution system - Detailed logging and trajectory recording for analysis - Parallel evaluation support for efficient benchmarking
Version History
- v1.0.0 (2025-08-07)
Initial public release of OmniEmbodied
Contributing to Changelog
When contributing to OmniEmbodied, please update this changelog:
Add entries to [Unreleased] section for future releases
Use appropriate categories: Added, Changed, Deprecated, Removed, Fixed, Security
Write clear, concise descriptions
Include issue/PR numbers where relevant
Follow the existing format
Changelog Categories:
Added - New features
Changed - Changes in existing functionality
Deprecated - Soon-to-be removed features
Removed - Features removed in this version
Fixed - Bug fixes
Security - Security vulnerability fixes
Release Schedule
OmniEmbodied follows semantic versioning:
Major versions (x.0.0) - Breaking changes, major new features
Minor versions (x.y.0) - New features, backward compatible
Patch versions (x.y.z) - Bug fixes, backward compatible
Support
For support questions, see Frequently Asked Questions or open an issue on GitHub.