Unity: A General Platform for Intelligent Agents

Abstract

Recent advances in artificial intelligence have been driven by the presence of increasingly realistic and complex simulated environments. However, many of the existing environments provide either unrealistic visuals, inaccurate physics, low task complexity, restricted agent perspective, or a limited capacity for interaction among artificial agents. Furthermore, many platforms lack the ability to flexibly configure the simulation, making the simulated environment a black-box from the perspective of the learning system. In this work, we propose a novel taxonomy of existing simulation platforms and discuss the highest level class of general platforms which enable the development of learning environments that are rich in visual, physical, task, and social complexity. We argue that modern game engines are uniquely suited to act as general platforms and as a case study examine the Unity engine and open source Unity ML-Agents Toolkit. We then survey the research enabled by Unity and the Unity ML-Agents Toolkit, discussing the kinds of research a flexible, interactive and easily configurable general platform can facilitate.

References in corpus (20)

• Playing Atari with Deep Reinforcement Learning
• Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor
• IMPALA: Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architectures
• DeepMind Control Suite
• Learning to reinforcement learn
• Emergent Tool Use From Multi-Agent Autocurricula
• AI2-THOR: An Interactive 3D Environment for Visual AI
• Leveraging Procedural Generation to Benchmark Reinforcement Learning
• Control of Memory, Active Perception, and Action in Minecraft
• Emergent Complexity via Multi-Agent Competition
• Paired Open-Ended Trailblazer (POET): Endlessly Generating Increasingly Complex and Diverse Learning Environments and Their Solutions
• Illuminating Generalization in Deep Reinforcement Learning through Procedural Level Generation
• Playing FPS Games with Deep Reinforcement Learning
• Gotta Learn Fast: A New Benchmark for Generalization in RL
• CHALET: Cornell House Agent Learning Environment
• Learning to Control Self-Assembling Morphologies: A Study of Generalization via Modularity
• Psychlab: A Psychology Laboratory for Deep Reinforcement Learning Agents
• AI-GAs: AI-generating algorithms, an alternate paradigm for producing general artificial intelligence
• On Learning to Think: Algorithmic Information Theory for Novel Combinations of Reinforcement Learning Controllers and Recurrent Neural World Models
• Winning Isn't Everything: Enhancing Game Development with Intelligent Agents

Cited by in corpus (76)

• A Survey and Critique of Multiagent Deep Reinforcement Learning
• Hybrid Reinforcement Learning-Based Eco-Driving Strategy for Connected and Automated Vehicles at Signalized Intersections
• A Survey of Deep Reinforcement Learning in Video Games
• Global Model Learning for Large Deformation Control of Elastic Deformable Linear Objects: An Efficient and Adaptive Approach
• Flightmare: A Flexible Quadrotor Simulator
• AllenAct: A Framework for Embodied AI Research
• Simulating Content Consistent Vehicle Datasets with Attribute Descent
• Predicting Game Difficulty and Churn Without Players
• Redirection Controller Using Reinforcement Learning
• Pegasus Simulator: An Isaac Sim Framework for Multiple Aerial Vehicles Simulation
• RLCard: A Toolkit for Reinforcement Learning in Card Games
• Gym-Ignition: Reproducible Robotic Simulations for Reinforcement Learning
• Potential Impacts of Smart Homes on Human Behavior: A Reinforcement Learning Approach
• Brax -- A Differentiable Physics Engine for Large Scale Rigid Body Simulation
• Target Search and Navigation in Heterogeneous Robot Systems with Deep Reinforcement Learning
• I Like to Move It: 6D Pose Estimation as an Action Decision Process
• A Domain-Agnostic Approach for Characterization of Lifelong Learning Systems
• Self-Supervised Exploration via Disagreement
• Strategies for Using Proximal Policy Optimization in Mobile Puzzle Games
• Archangel: A Hybrid UAV-based Human Detection Benchmark with Position and Pose Metadata
• Navigates Like Me: Understanding How People Evaluate Human-Like AI in Video Games
• 3DB: A Framework for Debugging Computer Vision Models
• Curriculum Learning for Safe Mapless Navigation
• RL-LABEL: A Deep Reinforcement Learning Approach Intended for AR Label Placement in Dynamic Scenarios
• Development of collective behavior in newborn artificial agents
• Ecological Reinforcement Learning
• Towards End-to-End Deep Learning for Autonomous Racing: On Data Collection and a Unified Architecture for Steering and Throttle Prediction
• Multi-Agent Reinforcement Learning with Multi-Step Generative Models
• A Deep Reinforcement Learning Approach to Audio-Based Navigation in a Multi-Speaker Environment
• ORRB -- OpenAI Remote Rendering Backend
• Cyber Mobility Mirror for Enabling Cooperative Driving Automation in Mixed Traffic: A Co-Simulation Platform
• Learning to Coordinate for a Worker-Station Multi-robot System in Planar Coverage Tasks
• NViSII: A Scriptable Tool for Photorealistic Image Generation
• Arena: A General Evaluation Platform and Building Toolkit for Multi-Agent Intelligence
• Fever Basketball: A Complex, Flexible, and Asynchronized Sports Game Environment for Multi-agent Reinforcement Learning
• Diver-Robot Communication Dataset for Underwater Hand Gesture Recognition
• Using Unity to Help Solve Intelligence
• Enhanced Adversarial Strategically-Timed Attacks against Deep Reinforcement Learning
• SLM Lab: A Comprehensive Benchmark and Modular Software Framework for Reproducible Deep Reinforcement Learning
• Marathon Environments: Multi-Agent Continuous Control Benchmarks in a Modern Video Game Engine
• Towards VEsNA, a Framework for Managing Virtual Environments via Natural Language Agents
• Learning to Engage with Interactive Systems: A Field Study on Deep Reinforcement Learning in a Public Museum
• Automated Play-Testing Through RL Based Human-Like Play-Styles Generation
• On the Use and Misuse of Absorbing States in Multi-agent Reinforcement Learning
• DeepCrawl: Deep Reinforcement Learning for Turn-based Strategy Games
• Central pattern generators evolved for real-time adaptation to rhythmic stimuli
• Hierarchical Control for Head-to-Head Autonomous Racing
• ARROCH: Augmented Reality for Robots Collaborating with a Human
• Learning to simulate complex scenes
• CUDA-Optimized real-time rendering of a Foveated Visual System
• Adaptive Synthetic Characters for Military Training
• Autonomous Curiosity for Real-Time Training Onboard Robotic Agents
• A GPU-based Hydrodynamic Simulator with Boid Interactions
• Structured agents for physical construction
• Fast Synthetic LiDAR Rendering via Spherical UV Unwrapping of Equirectangular Z-Buffer Images
• Learning Socially Appropriate Robot Approaching Behavior Toward Groups using Deep Reinforcement Learning
• Vision-driven UAV River Following: Benchmarking with Safe Reinforcement Learning
• Investigation of Factorized Optical Flows as Mid-Level Representations
• The AI Arena: A Framework for Distributed Multi-Agent Reinforcement Learning
• Meta Arcade: A Configurable Environment Suite for Meta-Learning
• Pessimistic Model Selection for Offline Deep Reinforcement Learning
• A Deep Reinforcement Learning Approach for Audio-based Navigation and Audio Source Localization in Multi-speaker Environments
• Benchmarking the Spectrum of Agent Capabilities
• Metis: Multi-Agent Based Crisis Simulation System
• Object-oriented state editing for HRL
• PPO Dash: Improving Generalization in Deep Reinforcement Learning
• Visual Perspective Taking for Opponent Behavior Modeling
• Learning to Grasp from 2.5D images: a Deep Reinforcement Learning Approach
• Robby is Not a Robber (anymore): On the Use of Institutions for Learning Normative Behavior
• Synthetic Data Are as Good as the Real for Association Knowledge Learning in Multi-object Tracking
• Proximal Policy Optimization with Mixed Distributed Training
• Game and Simulation Design for Studying Pedestrian-Automated Vehicle Interactions
• Meta Reinforcement Learning Based Sensor Scanning in 3D Uncertain Environments for Heterogeneous Multi-Robot Systems
• Fixed $\beta$-VAE Encoding for Curious Exploration in Complex 3D Environments
• Combine PPO with NES to Improve Exploration
• Wield: Systematic Reinforcement Learning With Progressive Randomization