Training and Simulation
R2026bTrain and simulate reinforcement learning agents
During training, the agent continuously updates its parameters to learn the optimal policy for a given environment. During simulation, the agent receives observations and a reward from the environment, and returns an action to the environment without updating its parameters.
Reinforcement Learning Toolbox™ provides functions for training agents and validating the training results through simulation. For an introduction to training and simulating agents, see Train Reinforcement Learning Agents.
Apps
| Reinforcement Learning Designer | Design, train, and simulate reinforcement learning agents |
Functions
Topics
Training and Simulation Basics
- Train Reinforcement Learning Agents
Find the optimal policy by training your agent within a specified environment. - Train Reinforcement Learning Agent in Basic Grid World
Train Q-learning and SARSA agents to solve a grid world in MATLAB®. - Train Reinforcement Learning Agent in MDP Environment
Train a reinforcement learning agent in a generic Markov decision process environment.
Use the Reinforcement Learning Designer App
- Specify Training Options in Reinforcement Learning Designer
Interactively specify options for training reinforcement learning agents using the Reinforcement Learning Designer app. - Specify Simulation Options in Reinforcement Learning Designer
Interactively specify options for simulating reinforcement learning agents using the Reinforcement Learning Designer app. - Design and Train Agent Using Reinforcement Learning Designer
Design and train a DQN agent for a cart-pole system using the Reinforcement Learning Designer app. - Design and Train Reinforcement Learning Agents Using Copilot Assisted Training Advisor
Use the AI-assisted Training Advisor to select and train a reinforcement learning agent. - Tune Hyperparameters Using Reinforcement Learning Designer
Search the hyperparameter space using Reinforcement Learning Designer.
Create and Train Default Agents
- Train Default DQN Agent to Balance Discrete Cart-Pole
Train a default DQN agent to balance discrete action space cart-pole system modeled in MATLAB. - Train Default DQN Agent to Swing Up and Balance Discrete Pendulum
Train a Default DQN agent to swing up and balance a discrete action space pendulum modeled in Simulink®. - Train Default DDPG Agent to Swing Up and Balance Continuous Pendulum
Train a DDPG agent to balance a continuous action space pendulum modeled in Simulink. - Train Default DDPG Agent to Swing Up and Balance Continuous Cart-Pole
Train a Default DDPG agent with custom networks to swing up and balance a continuous action space cart-pole system modeled in Simscape™ Multibody™. - Train Default PPO Agent for Discrete Lander Vehicle
Train a default PPO agent to land a discrete action space flying vehicle. - Train MBPO Agent to Balance Continuous Cart-Pole System
A model-based reinforcement learning agent learns a model of its environment and uses it to generate additional experiences for training.
Create and Train Agents with Custom Approximators
- Train LSPI Agent to Balance Discrete Cart-Pole
Train an LSPI agent to balance discrete action space cart-pole system modeled in MATLAB. - Train PG Agent with Custom Actor Network to Balance Discrete Cart-Pole
Train a PG agent with custom actor network to balance a discrete action space cart-pole system modeled in MATLAB. - Train PG Agent with Custom Actor and Baseline Networks to Control Discrete Double Integrator
Train a PG agent with a custom actor and baseline networks to control a discrete action space double integrator system modeled in MATLAB. - Train DDPG Agent with Custom Networks Using Image Observation
Train a DDPG agent with custom networks using an image-based observation signal. - Train Soft Actor Critic Agent with Custom Networks for Discrete Lander Vehicle
Train a SAC agent to land a discrete action space flying vehicle.
Create and Train Agents for Custom Simulink Environments
- Control Water Level in a Tank Using a DDPG Agent
Use DDPG agent to control water level in a tank modeled in Simulink. - Train DDPG Agent to Swing Up and Balance Pendulum with Bus Signal
Train a DDPG agent to balance a continuous action space pendulum Simulink model that contains observations in a bus signal.
Use Multiple Processes and GPUs
- Train Agents Using Parallel Computing and GPUs
Accelerate agent training by running simulations in parallel on multiple cores, GPUs, clusters or cloud resources. - Train AC Agent to Balance Discrete Cart-Pole Using Parallel Computing
Train an AC agent to control a discrete action space cart-pole system using asynchronous parallel computing. - Train DQN Agent for Lane Keeping Assist Using Parallel Computing
Train a DQN agent for an automated driving application using parallel computing.
Training and Simulation Advanced
- Train PPO Agent with Curriculum Learning for a Lane Keeping Application
Train a PPO agent for a lane keeping assist task by gradually increasing task complexity. - Train DQN Agent Using Hindsight Experience Replay
Train a DQN agent in a navigation environment with sparse rewards. - Train Reinforcement Learning Agent Offline to Control Quanser QUBE Pendulum
Train TD3 agent offline to control a Quanser QUBE pendulum. - Train Biped Robot to Walk Using Evolution Strategy-Reinforcement Learning Agents
Train TD3 agent using evolutionary strategy. - Create DQN Agent Using Deep Network Designer and Train Using Image Observations
Create a reinforcement learning agent using the Deep Network Designer app from the Deep Learning Toolbox™. - Transfer Learning: Fine-Tune DQN Agent for Pendulum Swing-Up from Earth to Mars
Use transfer learning to partially retrain a DQN agent to swing-up and balance a pendulum with Mars gravity conditions. - Enforce State-Dependent Action Constraints Using a Masking Network for a PPO Agent
Use an action-masking actor network within a PPO agent to prevent invalid actions.
Log Training Data and Tune Hyperparameters
- Log Training Data to Disk
Log a variety of data to disk while training an agent. - Train Agent or Tune Environment Parameters Using Parameter Sweeping
Tune a DDPG agent using hyperparameter sweeping. - Tune Hyperparameters Using Bayesian Optimization
Tune reinforcement learning hyperparameters using Bayesian optimization. - Configure Exploration for Reinforcement Learning Agents
Use visualization to configure exploration in reinforcement learning agents.
Multiagent Training
- Train Multiple Agents to Perform Collaborative Task
Train two continuous action space PPO agents to collaboratively move an object. - Train Multiple Agents for Area Coverage
Train three discrete action space PPO agents to explore a grid-world environment in a collaborative-competitive manner. - Train Multiple Agents for Path Following Control
Train a DQN and a DDPG agent to collaboratively perform adaptive cruise control and lane keeping assist to follow a path.
