Main Content

rlDQNAgent

Deep Q-network reinforcement learning agent

Description

The deep Q-network (DQN) algorithm is a model-free, online, off-policy reinforcement learning method. A DQN agent is a value-based reinforcement learning agent that trains a critic to estimate the return or future rewards. DQN is a variant of Q-learning, and it operates only within discrete action spaces.

For more information, Deep Q-Network Agents. For more information on the different types of reinforcement learning agents, see Reinforcement Learning Agents.

Creation

Description

Create Agent from Observation and Action Specifications

example

agent = rlDQNAgent(observationInfo,actionInfo) creates a DQN agent for an environment with the given observation and action specifications, using default initialization options. The critic in the agent uses a default vector (that is, multi-output) Q-value deep neural network built from the observation specification observationInfo and the action specification actionInfo. The ObservationInfo and ActionInfo properties of agent are set to the observationInfo and actionInfo input arguments, respectively.

example

agent = rlDQNAgent(observationInfo,actionInfo,initOpts) creates a DQN agent for an environment with the given observation and action specifications. The agent uses a default network configured using options specified in the initOpts object. For more information on the initialization options, see rlAgentInitializationOptions.

Create Agent from Critic

agent = rlDQNAgent(critic) creates a DQN agent with the specified critic network using a default option set for a DQN agent.

Specify Agent Options

example

agent = rlDQNAgent(critic,agentOptions) creates a DQN agent with the specified critic network and sets the AgentOptions property to the agentOptions input argument. Use this syntax after any of the input arguments in the previous syntaxes..

Input Arguments

expand all

Agent initialization options, specified as an rlAgentInitializationOptions object.

Critic, specified as an rlQValueFunction or as the generally more efficient rlVectorQValueFunction object. For more information on creating critics, see Create Policies and Value Functions.

Your critic can use a recurrent neural network as its function approximator. However, only rlVectorQValueFunction supports recurrent neural networks. For an example, see Create DQN Agent with Recurrent Neural Network.

Properties

expand all

Observation specifications, specified as a reinforcement learning specification object or an array of specification objects defining properties such as dimensions, data type, and names of the observation signals.

If you create the agent by specifying a critic object, the value of ObservationInfo matches the value specified in critic.

You can extract observationInfo from an existing environment or agent using getObservationInfo. You can also construct the specifications manually using rlFiniteSetSpec or rlNumericSpec.

Action specifications, specified as a reinforcement learning specification object defining properties such as dimensions, data type, and names of the action signals.

Since a DDPG agent operates in a discrete action space, you must specify actionInfo as an rlFiniteSetSpec object.

If you create the agent by specifying a critic object, the value of ActionInfo matches the value specified in critic.

You can extract actionInfo from an existing environment or agent using getActionInfo. You can also construct the specification manually using rlFiniteSetSpec.

Agent options, specified as an rlDQNAgentOptions object.

If you create a DQN agent with a default critic that uses a recurrent neural network, the default value of AgentOptions.SequenceLength is 32.

Experience buffer, specified as an rlReplayMemory object. During training the agent stores each of its experiences (S,A,R,S',D) in a buffer. Here:

  • S is the current observation of the environment.

  • A is the action taken by the agent.

  • R is the reward for taking action A.

  • S' is the next observation after taking action A.

  • D is the is-done signal after taking action A.

Option to use exploration policy when selecting actions, specified as a one of the following logical values.

  • true — Use the base agent exploration policy when selecting actions.

  • false — Use the base agent greedy policy when selecting actions.

Sample time of agent, specified as a positive scalar or as -1. Setting this parameter to -1 allows for event-based simulations. The value of SampleTime matches the value specified in AgentOptions.

Within a Simulink® environment, the RL Agent block in which the agent is specified to execute every SampleTime seconds of simulation time. If SampleTime is -1, the block inherits the sample time from its parent subsystem.

Within a MATLAB® environment, the agent is executed every time the environment advances. In this case, SampleTime is the time interval between consecutive elements in the output experience returned by sim or train. If SampleTime is -1, the time interval between consecutive elements in the returned output experience reflects the timing of the event that triggers the agent execution.

Object Functions

trainTrain reinforcement learning agents within a specified environment
simSimulate trained reinforcement learning agents within specified environment
getActionObtain action from agent or actor given environment observations
getActorGet actor from reinforcement learning agent
setActorSet actor of reinforcement learning agent
getCriticGet critic from reinforcement learning agent
setCriticSet critic of reinforcement learning agent
generatePolicyFunctionCreate function that evaluates trained policy of reinforcement learning agent

Examples

collapse all

Create an environment with a discrete action space, and obtain its observation and action specifications. For this example, load the environment used in the example Create Agent Using Deep Network Designer and Train Using Image Observations. This environment has two observations: a 50-by-50 grayscale image and a scalar (the angular velocity of the pendulum). The action is a scalar with five possible elements (a torque of either -2, -1, 0, 1, or 2 Nm applied to a swinging pole).

% load predefined environment
env = rlPredefinedEnv("SimplePendulumWithImage-Discrete");

% obtain observation and action specifications
obsInfo = getObservationInfo(env);
actInfo = getActionInfo(env);

The agent creation function initializes the actor and critic networks randomly. You can ensure reproducibility by fixing the seed of the random generator.

rng(0)

Create a deep Q-network agent from the environment observation and action specifications.

agent = rlDQNAgent(obsInfo,actInfo);

To check your agent, use getAction to return the action from a random observation.

getAction(agent,{rand(obsInfo(1).Dimension),rand(obsInfo(2).Dimension)})
ans = 1x1 cell array
    {[1]}

You can now test and train the agent within the environment.

Create an environment with a discrete action space, and obtain its observation and action specifications. For this example, load the environment used in the example Create Agent Using Deep Network Designer and Train Using Image Observations. This environment has two observations: a 50-by-50 grayscale image and a scalar (the angular velocity of the pendulum). The action is a scalar with five possible elements (a torque of either -2, -1, 0, 1, or 2 Nm applied to a swinging pole).

% load predefined environment
env = rlPredefinedEnv("SimplePendulumWithImage-Discrete");

% obtain observation and action specifications
obsInfo = getObservationInfo(env);
actInfo = getActionInfo(env);

Create an agent initialization option object, specifying that each hidden fully connected layer in the network must have 128 neurons (instead of the default number, 256).

initOpts = rlAgentInitializationOptions('NumHiddenUnit',128);

The agent creation function initializes the actor and critic networks randomly. You can ensure reproducibility by fixing the seed of the random generator. To do so, uncomment the following line.

% rng(0)

Create a policy gradient agent from the environment observation and action specifications.

agent = rlPGAgent(obsInfo,actInfo,initOpts);

Extract the deep neural network from both the critic.

criticNet = getModel(getCritic(agent));

The default DQN agent uses a multi-output Q-value critic approximator. A multi-output approximator has observations as inputs and state-action values as outputs. Each output element represents the expected cumulative long-term reward for taking the corresponding discrete action from the state indicated by the observation inputs.

Display the layers of the critic network, and verify that each hidden fully connected layer has 128 neurons

criticNet.Layers
ans = 
  11x1 Layer array with layers:

     1   'concat'         Concatenation     Concatenation of 2 inputs along dimension 1
     2   'relu_body'      ReLU              ReLU
     3   'fc_body'        Fully Connected   128 fully connected layer
     4   'body_output'    ReLU              ReLU
     5   'input_1'        Image Input       50x50x1 images
     6   'conv_1'         Convolution       64 3x3x1 convolutions with stride [1  1] and padding [0  0  0  0]
     7   'relu_input_1'   ReLU              ReLU
     8   'fc_1'           Fully Connected   128 fully connected layer
     9   'input_2'        Feature Input     1 features
    10   'fc_2'           Fully Connected   128 fully connected layer
    11   'output'         Fully Connected   1 fully connected layer

Plot the critic network

plot(layerGraph(criticNet))

Figure contains an axes object. The axes object contains an object of type graphplot.

To check your agent, use getAction to return the action from a random observation.

getAction(agent,{rand(obsInfo(1).Dimension),rand(obsInfo(2).Dimension)})
ans = 1x1 cell array
    {[2]}

You can now test and train the agent within the environment.

Create an environment interface and obtain its observation and action specifications. For this example load the predefined environment used for the Train DQN Agent to Balance Cart-Pole System example. This environment has a continuous four-dimensional observation space (the positions and velocities of both cart and pole) and a discrete one-dimensional action space consisting on the application of two possible forces, -10N or 10N.

% load predefined environment
env = rlPredefinedEnv("CartPole-Discrete");

% get observation and action specification objects
obsInfo = getObservationInfo(env);
actInfo = getActionInfo(env);

Create a deep neural network to be used as approximation model within the critic. For DQN agents, you have the option to create a multi-output Q-value function critic, which is generally more efficient than a comparable single-output critic.

The network for this critic must take only the observation as input and return a vector of values for each action. Therefore, it must have an input layer with as many elements as the dimension of the observation space and an output layer having as many elements as the number of possible discrete actions. Each output element represents the expected cumulative long-term reward following from the observation given as input, when the corresponding action is taken.

dnn = [
    featureInputLayer(prod(obsInfo.Dimension), ...
        'Normalization', 'none', 'Name', 'state')
    fullyConnectedLayer(24, 'Name', 'CriticStateFC1')
    reluLayer('Name', 'CriticRelu1')
    fullyConnectedLayer(24, 'Name', 'CriticStateFC2')
    reluLayer('Name','CriticCommonRelu')
    fullyConnectedLayer(length(actInfo.Elements), ...
        'Name', 'output')];

Create the critic using rlVectorQValueFunction, the network dnn as well as the observation and action specifications.

critic = rlVectorQValueFunction(dnn,obsInfo,actInfo);

Define some training options for the critic.

criticOpts = rlOptimizerOptions( ...
    'LearnRate',1e-2,'GradientThreshold',1);

Specify agent options, and create a DQN agent using the critic.

agentOpts = rlDQNAgentOptions(...
    'UseDoubleDQN',false, ...    
    'TargetUpdateMethod',"periodic", ...
    'TargetUpdateFrequency',4, ...   
    'ExperienceBufferLength',100000, ...
    'DiscountFactor',0.99, ...
    'MiniBatchSize',256, ...
    'CriticOptimizerOptions',criticOpts);
agent = rlDQNAgent(critic,agentOpts)
agent = 
  rlDQNAgent with properties:

        ExperienceBuffer: [1x1 rl.replay.rlReplayMemory]
            AgentOptions: [1x1 rl.option.rlDQNAgentOptions]
    UseExplorationPolicy: 0
         ObservationInfo: [1x1 rl.util.rlNumericSpec]
              ActionInfo: [1x1 rl.util.rlFiniteSetSpec]
              SampleTime: 1

To check your agent, use getAction to return the action from a random observation.

getAction(agent,{rand(obsInfo.Dimension)})
ans = 1x1 cell array
    {[10]}

You can now test and train the agent against the environment.

Create an environment interface and obtain its observation and action specifications. For this example load the predefined environment used for the Train DQN Agent to Balance Cart-Pole System example. This environment has a continuous four-dimensional observation space (the positions and velocities of both cart and pole) and a discrete one-dimensional action space consisting on the application of two possible forces, -10N or 10N.

env = rlPredefinedEnv("CartPole-Discrete");
obsInfo = getObservationInfo(env);
actInfo = getActionInfo(env);

Create a deep neural network to be used as approximation model within the critic. For DQN agents, you have the option to create a multi-output Q-value function critic, which is generally more efficient than a comparable single-output critic. However, for this example, create a single-output Q-value function critic instead.

The network for this critic must have two input layers, one for the observation and the other for the action, and return a scalar value representing the expected cumulative long-term reward following from the given observation and action.

% define paths
statePath = [
    featureInputLayer(prod(obsInfo.Dimension), ...
        'Normalization', 'none', 'Name', 'state')
    fullyConnectedLayer(24, 'Name', 'CriticStateFC1')
    reluLayer('Name', 'CriticRelu1')
    fullyConnectedLayer(24, 'Name', 'CriticStateFC2')];
actionPath = [
    featureInputLayer(prod(actInfo.Dimension), ...
         'Normalization', 'none', 'Name', 'action')
    fullyConnectedLayer(24, 'Name', 'CriticActionFC1')];
commonPath = [
    additionLayer(2,'Name', 'add')
    reluLayer('Name','CriticCommonRelu')
    fullyConnectedLayer(1, 'Name', 'output')];

% add paths to network
criticNetwork = layerGraph(statePath);
criticNetwork = addLayers(criticNetwork, actionPath);
criticNetwork = addLayers(criticNetwork, commonPath);

% connect layers
criticNetwork = connectLayers(criticNetwork, ...
    'CriticStateFC2','add/in1');
criticNetwork = connectLayers(criticNetwork, ...
    'CriticActionFC1','add/in2');

Create the critic using rlQValueFunction, criticNetwork, as well as the observation and action specifications.

critic = rlQValueFunction(criticNetwork,obsInfo,actInfo);

Define some training options for the critic.

criticOptions = rlOptimizerOptions( ...
    'LearnRate',1e-2,'GradientThreshold',1);

Specify agent options, and create a DQN agent using the critic.

agentOpts = rlDQNAgentOptions(...
    'UseDoubleDQN',false, ...    
    'TargetUpdateMethod',"periodic", ...
    'TargetUpdateFrequency',4, ...   
    'ExperienceBufferLength',100000, ...
    'DiscountFactor',0.99, ...
    'MiniBatchSize',256, ...
    'CriticOptimizerOptions',criticOptions);

agent = rlDQNAgent(critic,agentOpts)
agent = 
  rlDQNAgent with properties:

        ExperienceBuffer: [1x1 rl.replay.rlReplayMemory]
            AgentOptions: [1x1 rl.option.rlDQNAgentOptions]
    UseExplorationPolicy: 0
         ObservationInfo: [1x1 rl.util.rlNumericSpec]
              ActionInfo: [1x1 rl.util.rlFiniteSetSpec]
              SampleTime: 1

To check your agent, use getAction to return the action from a random observation.

getAction(agent,{rand(4,1)})
ans = 1x1 cell array
    {[10]}

You can now test and train the agent against the environment.

For this example load the predefined environment used for the Train DQN Agent to Balance Cart-Pole System example.

env = rlPredefinedEnv('CartPole-Discrete');

Get observation and action information. This environment has a continuous four-dimensional observation space (the positions and velocities of both cart and pole) and a discrete one-dimensional action space consisting on the application of two possible forces, -10N or 10N.

obsInfo = getObservationInfo(env);
actInfo = getActionInfo(env);

Create a recurrent deep neural network for your critic. To create a recurrent neural network, use a sequenceInputLayer as the input layer and include an lstmLayer as one of the other network layers.

For DQN agents, only the vector function approximator, rlVectorQValueFunction, supports recurrent neural networks models. For vector Q-value function critics, the number of elements of the output layer has to be equal to the number of possible actions: numel(actInfo.Elements).

criticNetwork = [
    sequenceInputLayer(obsInfo.Dimension(1), ...
        'Normalization','none','Name','state')
    fullyConnectedLayer(50, 'Name', 'cFC1')
    reluLayer('Name','cRelu1')
    lstmLayer(20,'OutputMode','sequence','Name','cLSTM');
    fullyConnectedLayer(20,'Name','cFC2')
    reluLayer('Name','cRelu2')
    fullyConnectedLayer(numel(actInfo.Elements), ...
        'Name','output')];

Create a critic using the recurrent neural network.

critic = rlVectorQValueFunction(criticNetwork,obsInfo,actInfo);

Define some training options for the critic.

criticOptions = rlOptimizerOptions( ...
    'LearnRate',1e-3,'GradientThreshold',1);

Specify options for creating the DQN agent. To use a recurrent neural network, you must specify a SequenceLength greater than 1.

agentOptions = rlDQNAgentOptions(...
    'UseDoubleDQN',false, ...
    'TargetSmoothFactor',5e-3, ...
    'ExperienceBufferLength',1e6, ...
    'SequenceLength',32, ...
    'CriticOptimizerOptions',criticOptions);

agentOptions.EpsilonGreedyExploration.EpsilonDecay = 1e-4;

Create the agent. The actor and critic networks are initialized randomly.

agent = rlDQNAgent(critic,agentOptions);

Check your agent using getAction to return the action from a random observation.

getAction(agent,rand(obsInfo.Dimension))
ans = 1x1 cell array
    {[-10]}

You can now test and train the agent against the environment.

Version History

Introduced in R2019a