PLATFORMS FOR THE DEVELOPMENT OF AGENT SYSTEMS

Platforms for Artificial Intelligence Research

GO BACK

This section includes platforms used in artificial intelligence research. They are mainly used for simulating environments that are later used for learning purposes for AI controlled agents.

# Name Last version Programming language License Website, Documentation, Papers Description Status
1 Brax 0.9.2 Python
Physics engine containing algorithms for agents training used in robotics research ACTIVE
2 Deepmind Garage 2020.06.3 Python
Toolkit that facilitates reinforcement learning algorithms and provides a standardized, reproducible environment ACTIVE
3 Deepmind Lab December 2020 C, Lua
3D game platform used for exploring learning agents, which aims to serve as a testing environment in AI research ACTIVE
4 Gazebo Garden C++
Cross-platform 3D robotics simulator that facilitates high-fidelity physics, rendering, and sensor models ACTIVE
5 Gymnasium 0.29.1 Python
Toolkit for developing reinforcement learning algorithms that facilitate API for single-agent RL ACTIVE
6 Habitat 0.2.5 Python
Simulation platform that enables training Embodied AI agents and offers high-performance 3D simulator ACTIVE
7 MAgent2 0.3.2 Python
Engine for generating high-performance multi-agent grid environments with many agents ACTIVE
8 Mava 0.1.3 Python
JAX-based framework for multi-agent reinforcement learning systems ACTIVE
9 MuJuCo 2.3.7 C, C++, C#, Python
General purpose physics engine with native 3D modelling environment that facilitates fast and reliable simulations ACTIVE
10 Open Spiel 1.3 C++, Python
Collection of environments and algorithms for research in general reinforcement learning and search/planning in games ACTIVE
11 PySC2 4.0 Python
Reinforcement learning Environment inspired by StarCraft II that involves multi-agent interactions ACTIVE
12 RecSim 0.2.4 Python
Platform for simulation of environments for recommender systems, supports development of new recommendation algorithms UNCLEAR
13 Unity ML-agents 20 C#, Python
Platform that facilitates the training of intelligent agents that is dedicated for game development and AI research ACTIVE