using System; using System.Collections.Generic; using UnityEngine; using Barracuda; using MLAgents.Sensor; using UnityEngine.Serialization; namespace MLAgents { /// /// Struct that contains all the information for an Agent, including its /// observations, actions and current status, that is sent to the Brain. /// public struct AgentInfo { /// /// Most recent observations. /// public List observations; /// /// Keeps track of the last vector action taken by the Brain. /// public float[] storedVectorActions; /// /// For discrete control, specifies the actions that the agent cannot take. Is true if /// the action is masked. /// public bool[] actionMasks; /// /// Current agent reward. /// public float reward; /// /// Whether the agent is done or not. /// public bool done; /// /// Whether the agent has reached its max step count for this episode. /// public bool maxStepReached; /// /// Unique identifier each agent receives at initialization. It is used /// to separate between different agents in the environment. /// public int id; } /// /// Struct that contains the action information sent from the Brain to the /// Agent. /// public struct AgentAction { public float[] vectorActions; public float value; } /// /// Struct that contains all the Agent-specific parameters provided in the /// Editor. This excludes the Brain linked to the Agent since it can be /// modified programmatically. /// [Serializable] public class AgentParameters { /// /// The maximum number of steps the agent takes before being done. /// /// /// If set to 0, the agent can only be set to done programmatically (or /// when the Academy is done). /// If set to any positive integer, the agent will be set to done after /// that many steps. Note that setting the max step to a value greater /// than the academy max step value renders it useless. /// public int maxStep; /// /// Determines the behaviour of the agent when done. /// /// /// If true, the agent will reset when done and start a new episode. /// Otherwise, the agent will remain done and its behavior will be /// dictated by the AgentOnDone method. /// public bool resetOnDone = true; /// /// Whether to enable On Demand Decisions or make a decision at /// every step. /// public bool onDemandDecision; /// /// Number of actions between decisions (used when On Demand Decisions /// is turned off). /// public int numberOfActionsBetweenDecisions; } /// /// Agent Monobehavior class that is attached to a Unity GameObject, making it /// an Agent. An agent produces observations and takes actions in the /// environment. Observations are determined by the cameras attached /// to the agent in addition to the vector observations implemented by the /// user in . On the other hand, actions /// are determined by decisions produced by a Policy. Currently, this /// class is expected to be extended to implement the desired agent behavior. /// /// /// Simply speaking, an agent roams through an environment and at each step /// of the environment extracts its current observation, sends them to its /// policy and in return receives an action. In practice, /// however, an agent need not send its observation at every step since very /// little may have changed between successive steps. /// /// At any step, an agent may be considered . /// This could occur due to a variety of reasons: /// - The agent reached an end state within its environment. /// - The agent reached the maximum # of steps (i.e. timed out). /// - The academy reached the maximum # of steps (forced agent to be done). /// /// Here, an agent reaches an end state if it completes its task successfully /// or somehow fails along the way. In the case where an agent is done before /// the academy, it either resets and restarts, or just lingers until the /// academy is done. /// /// An important note regarding steps and episodes is due. Here, an agent step /// corresponds to an academy step, which also corresponds to Unity /// environment step (i.e. each FixedUpdate call). This is not the case for /// episodes. The academy controls the global episode count and each agent /// controls its own local episode count and can reset and start a new local /// episode independently (based on its own experience). Thus an academy /// (global) episode can be viewed as the upper-bound on an agents episode /// length and that within a single global episode, an agent may have completed /// multiple local episodes. Consequently, if an agent max step is /// set to a value larger than the academy max steps value, then the academy /// value takes precedence (since the agent max step will never be reached). /// /// Lastly, note that at any step the policy to the agent is allowed to /// change model with . /// /// Implementation-wise, it is required that this class is extended and the /// virtual methods overridden. For sample implementations of agent behavior, /// see the Examples/ directory within this Unity project. /// [HelpURL("https://github.com/Unity-Technologies/ml-agents/blob/master/" + "docs/Learning-Environment-Design-Agents.md")] [Serializable] [RequireComponent(typeof(BehaviorParameters))] public abstract class Agent : MonoBehaviour { IPolicy m_Brain; BehaviorParameters m_PolicyFactory; /// /// Agent parameters specified within the Editor via AgentEditor. /// [HideInInspector] public AgentParameters agentParameters; /// Current Agent information (message sent to Brain). AgentInfo m_Info; public AgentInfo Info { get { return m_Info; } set { m_Info = value; } } /// Current Agent action (message sent from Brain). AgentAction m_Action; /// Represents the reward the agent accumulated during the current step. /// It is reset to 0 at the beginning of every step. /// Should be set to a positive value when the agent performs a "good" /// action that we wish to reinforce/reward, and set to a negative value /// when the agent performs a "bad" action that we wish to punish/deter. /// Additionally, the magnitude of the reward should not exceed 1.0 float m_Reward; /// Keeps track of the cumulative reward in this episode. float m_CumulativeReward; /// Whether or not the agent requests an action. bool m_RequestAction; /// Whether or not the agent requests a decision. bool m_RequestDecision; /// Whether or not the agent has completed the episode. This may be due /// to either reaching a success or fail state, or reaching the maximum /// number of steps (i.e. timing out). bool m_Done; /// Whether or not the agent reached the maximum number of steps. bool m_MaxStepReached; /// Keeps track of the number of steps taken by the agent in this episode. /// Note that this value is different for each agent, and may not overlap /// with the step counter in the Academy, since agents reset based on /// their own experience. int m_StepCount; /// Flag to signify that an agent has been reset but the fact that it is /// done has not been communicated (required for On Demand Decisions). bool m_HasAlreadyReset; /// Flag to signify that an agent is done and should not reset until /// the fact that it is done has been communicated. bool m_Terminate; /// Unique identifier each agent receives at initialization. It is used /// to separate between different agents in the environment. int m_Id; /// Keeps track of the actions that are masked at each step. ActionMasker m_ActionMasker; /// /// Demonstration recorder. /// DemonstrationRecorder m_Recorder; /// /// List of sensors used to generate observations. /// Currently generated from attached SensorComponents, and a legacy VectorSensor /// [FormerlySerializedAs("m_Sensors")] public List sensors; /// /// VectorSensor which is written to by AddVectorObs /// public VectorSensor collectObservationsSensor; /// /// Internal buffer used for generating float observations. /// float[] m_VectorSensorBuffer; WriteAdapter m_WriteAdapter = new WriteAdapter(); /// MonoBehaviour function that is called when the attached GameObject /// becomes enabled or active. void OnEnable() { m_Id = gameObject.GetInstanceID(); var academy = FindObjectOfType(); academy.LazyInitialization(); OnEnableHelper(academy); m_Recorder = GetComponent(); } /// Helper method for the event, created to /// facilitate testing. void OnEnableHelper(Academy academy) { m_Info = new AgentInfo(); m_Action = new AgentAction(); sensors = new List(); if (academy == null) { throw new UnityAgentsException( "No Academy Component could be found in the scene."); } academy.AgentSetStatus += SetStatus; academy.AgentResetIfDone += ResetIfDone; academy.AgentSendState += SendInfo; academy.DecideAction += DecideAction; academy.AgentAct += AgentStep; academy.AgentForceReset += _AgentReset; m_PolicyFactory = GetComponent(); m_Brain = m_PolicyFactory.GeneratePolicy(Heuristic); ResetData(); InitializeAgent(); InitializeSensors(); } /// Monobehavior function that is called when the attached GameObject /// becomes disabled or inactive. void OnDisable() { var academy = FindObjectOfType(); if (academy != null) { academy.AgentSetStatus -= SetStatus; academy.AgentResetIfDone -= ResetIfDone; academy.AgentSendState -= SendInfo; academy.DecideAction -= DecideAction; academy.AgentAct -= AgentStep; academy.AgentForceReset -= ForceReset; } m_Brain?.Dispose(); } /// /// Updates the Model for the agent. Any model currently assigned to the /// agent will be replaced with the provided one. If the arguments are /// identical to the current parameters of the agent, the model will /// remain unchanged. /// /// The identifier of the behavior. This /// will categorize the agent when training. /// /// The model to use for inference. /// Define on what device the model /// will be run. public void GiveModel( string behaviorName, NNModel model, InferenceDevice inferenceDevice = InferenceDevice.CPU) { m_PolicyFactory.GiveModel(behaviorName, model, inferenceDevice); m_Brain?.Dispose(); m_Brain = m_PolicyFactory.GeneratePolicy(Heuristic); } /// /// Returns the current step counter (within the current epside). /// /// /// Current episode number. /// public int GetStepCount() { return m_StepCount; } /// /// Resets the step reward and possibly the episode reward for the agent. /// public void ResetReward() { m_Reward = 0f; if (m_Done) { m_CumulativeReward = 0f; } } /// /// Overrides the current step reward of the agent and updates the episode /// reward accordingly. /// /// The new value of the reward. public void SetReward(float reward) { m_CumulativeReward += (reward - m_Reward); m_Reward = reward; } /// /// Increments the step and episode rewards by the provided value. /// /// Incremental reward value. public void AddReward(float increment) { m_Reward += increment; m_CumulativeReward += increment; } /// /// Retrieves the step reward for the Agent. /// /// The step reward. public float GetReward() { return m_Reward; } /// /// Retrieves the episode reward for the Agent. /// /// The episode reward. public float GetCumulativeReward() { return m_CumulativeReward; } /// /// Sets the done flag to true. /// public void Done() { m_Done = true; } /// /// Is called when the agent must request the brain for a new decision. /// public void RequestDecision() { m_RequestDecision = true; RequestAction(); } /// /// Is called then the agent must perform a new action. /// public void RequestAction() { m_RequestAction = true; } /// /// Indicates if the agent has reached his maximum number of steps. /// /// /// true, if max step reached was reached, false otherwise. /// public bool IsMaxStepReached() { return m_MaxStepReached; } /// /// Indicates if the agent is done /// /// /// true, if the agent is done, false otherwise. /// public bool IsDone() { return m_Done; } /// Helper function that resets all the data structures associated with /// the agent. Typically used when the agent is being initialized or reset /// at the end of an episode. void ResetData() { var param = m_PolicyFactory.brainParameters; m_ActionMasker = new ActionMasker(param); // If we haven't initialized vectorActions, initialize to 0. This should only // happen during the creation of the Agent. In subsequent episodes, vectorAction // should stay the previous action before the Done(), so that it is properly recorded. if (m_Action.vectorActions == null) { if (param.vectorActionSpaceType == SpaceType.Continuous) { m_Action.vectorActions = new float[param.vectorActionSize[0]]; m_Info.storedVectorActions = new float[param.vectorActionSize[0]]; } else { m_Action.vectorActions = new float[param.vectorActionSize.Length]; m_Info.storedVectorActions = new float[param.vectorActionSize.Length]; } } m_Info.observations = new List(); } /// /// Initializes the agent, called once when the agent is enabled. Can be /// left empty if there is no special, unique set-up behavior for the /// agent. /// /// /// One sample use is to store local references to other objects in the /// scene which would facilitate computing this agents observation. /// public virtual void InitializeAgent() { } /// /// When the Agent uses Heuristics, it will call this method every time it /// needs an action. This can be used for debugging or controlling the agent /// with keyboard. /// /// A float array corresponding to the next action of the Agent /// public virtual float[] Heuristic() { throw new UnityAgentsException(string.Format( "The Heuristic method was not implemented for the Agent on the " + "{0} GameObject.", gameObject.name)); } /// /// Set up the list of ISensors on the Agent. By default, this will select any /// SensorBase's attached to the Agent. /// public void InitializeSensors() { // Get all attached sensor components var attachedSensorComponents = GetComponents(); sensors.Capacity += attachedSensorComponents.Length; foreach (var component in attachedSensorComponents) { sensors.Add(component.CreateSensor()); } // Support legacy CollectObservations var param = m_PolicyFactory.brainParameters; if (param.vectorObservationSize > 0) { collectObservationsSensor = new VectorSensor(param.vectorObservationSize); if (param.numStackedVectorObservations > 1) { var stackingSensor = new StackingSensor(collectObservationsSensor, param.numStackedVectorObservations); sensors.Add(stackingSensor); } else { sensors.Add(collectObservationsSensor); } } // Sort the Sensors by name to ensure determinism sensors.Sort((x, y) => x.GetName().CompareTo(y.GetName())); #if DEBUG // Make sure the names are actually unique for (var i = 0; i < sensors.Count - 1; i++) { Debug.Assert(!sensors[i].GetName().Equals(sensors[i + 1].GetName()), "Sensor names must be unique."); } #endif // Create a buffer for writing vector sensor data too int numFloatObservations = 0; for (var i = 0; i < sensors.Count; i++) { if (sensors[i].GetCompressionType() == SensorCompressionType.None) { numFloatObservations += sensors[i].ObservationSize(); } } m_VectorSensorBuffer = new float[numFloatObservations]; } /// /// Sends the Agent info to the linked Brain. /// void SendInfoToBrain() { if (m_Brain == null) { return; } m_Info.storedVectorActions = m_Action.vectorActions; m_Info.observations.Clear(); m_ActionMasker.ResetMask(); UpdateSensors(); using (TimerStack.Instance.Scoped("CollectObservations")) { CollectObservations(); } m_Info.actionMasks = m_ActionMasker.GetMask(); // var param = m_PolicyFactory.brainParameters; // look, no brain params! m_Info.reward = m_Reward; m_Info.done = m_Done; m_Info.maxStepReached = m_MaxStepReached; m_Info.id = m_Id; m_Brain.RequestDecision(this); if (m_Recorder != null && m_Recorder.record && Application.isEditor) { // This is a bit of a hack - if we're in inference mode, observations won't be generated // But we need these to be generated for the recorder. So generate them here. if (m_Info.observations.Count == 0) { GenerateSensorData(); } m_Recorder.WriteExperience(m_Info); } } void UpdateSensors() { for (var i = 0; i < sensors.Count; i++) { sensors[i].Update(); } } /// /// Generate data for each sensor and store it on the Agent's AgentInfo. /// NOTE: At the moment, this is only called during training or when using a DemonstrationRecorder; /// during inference the Sensors are used to write directly to the Tensor data. This will likely change in the /// future to be controlled by the type of brain being used. /// public void GenerateSensorData() { int floatsWritten = 0; // Generate data for all Sensors for (var i = 0; i < sensors.Count; i++) { var sensor = sensors[i]; if (sensor.GetCompressionType() == SensorCompressionType.None) { // only handles 1D // TODO handle in communicator code instead m_WriteAdapter.SetTarget(m_VectorSensorBuffer, floatsWritten); var numFloats = sensor.Write(m_WriteAdapter); var floatObs = new Observation { FloatData = new ArraySegment(m_VectorSensorBuffer, floatsWritten, numFloats), Shape = sensor.GetFloatObservationShape(), CompressionType = sensor.GetCompressionType() }; m_Info.observations.Add(floatObs); floatsWritten += numFloats; } else { var compressedObs = new Observation { CompressedData = sensor.GetCompressedObservation(), Shape = sensor.GetFloatObservationShape(), CompressionType = sensor.GetCompressionType() }; m_Info.observations.Add(compressedObs); } } } /// /// Collects the (vector, visual) observations of the agent. /// The agent observation describes the current environment from the /// perspective of the agent. /// /// /// Simply, an agents observation is any environment information that helps /// the Agent acheive its goal. For example, for a fighting Agent, its /// observation could include distances to friends or enemies, or the /// current level of ammunition at its disposal. /// Recall that an Agent may attach vector or visual observations. /// Vector observations are added by calling the provided helper methods: /// - /// - /// - /// - /// - /// AddVectorObs(float[]) /// /// - /// AddVectorObs(List{float}) /// /// - /// - /// - /// Depending on your environment, any combination of these helpers can /// be used. They just need to be used in the exact same order each time /// this method is called and the resulting size of the vector observation /// needs to match the vectorObservationSize attribute of the linked Brain. /// Visual observations are implicitly added from the cameras attached to /// the Agent. /// public virtual void CollectObservations() { } /// /// Sets an action mask for discrete control agents. When used, the agent will not be /// able to perform the action passed as argument at the next decision. If no branch is /// specified, the default branch will be 0. The actionIndex or actionIndices correspond /// to the action the agent will be unable to perform. /// /// The indices of the masked actions on branch 0 protected void SetActionMask(IEnumerable actionIndices) { m_ActionMasker.SetActionMask(0, actionIndices); } /// /// Sets an action mask for discrete control agents. When used, the agent will not be /// able to perform the action passed as argument at the next decision. If no branch is /// specified, the default branch will be 0. The actionIndex or actionIndices correspond /// to the action the agent will be unable to perform. /// /// The index of the masked action on branch 0 protected void SetActionMask(int actionIndex) { m_ActionMasker.SetActionMask(0, new[] { actionIndex }); } /// /// Sets an action mask for discrete control agents. When used, the agent will not be /// able to perform the action passed as argument at the next decision. If no branch is /// specified, the default branch will be 0. The actionIndex or actionIndices correspond /// to the action the agent will be unable to perform. /// /// The branch for which the actions will be masked /// The index of the masked action protected void SetActionMask(int branch, int actionIndex) { m_ActionMasker.SetActionMask(branch, new[] { actionIndex }); } /// /// Modifies an action mask for discrete control agents. When used, the agent will not be /// able to perform the action passed as argument at the next decision. If no branch is /// specified, the default branch will be 0. The actionIndex or actionIndices correspond /// to the action the agent will be unable to perform. /// /// The branch for which the actions will be masked /// The indices of the masked actions protected void SetActionMask(int branch, IEnumerable actionIndices) { m_ActionMasker.SetActionMask(branch, actionIndices); } /// /// Adds a float observation to the vector observations of the agent. /// Increases the size of the agents vector observation by 1. /// /// Observation. protected void AddVectorObs(float observation) { collectObservationsSensor.AddObservation(observation); } /// /// Adds an integer observation to the vector observations of the agent. /// Increases the size of the agents vector observation by 1. /// /// Observation. protected void AddVectorObs(int observation) { collectObservationsSensor.AddObservation(observation); } /// /// Adds an Vector3 observation to the vector observations of the agent. /// Increases the size of the agents vector observation by 3. /// /// Observation. protected void AddVectorObs(Vector3 observation) { collectObservationsSensor.AddObservation(observation); } /// /// Adds an Vector2 observation to the vector observations of the agent. /// Increases the size of the agents vector observation by 2. /// /// Observation. protected void AddVectorObs(Vector2 observation) { collectObservationsSensor.AddObservation(observation); } /// /// Adds a collection of float observations to the vector observations of the agent. /// Increases the size of the agents vector observation by size of the collection. /// /// Observation. protected void AddVectorObs(IEnumerable observation) { collectObservationsSensor.AddObservation(observation); } /// /// Adds a quaternion observation to the vector observations of the agent. /// Increases the size of the agents vector observation by 4. /// /// Observation. protected void AddVectorObs(Quaternion observation) { collectObservationsSensor.AddObservation(observation); } /// /// Adds a boolean observation to the vector observation of the agent. /// Increases the size of the agent's vector observation by 1. /// /// protected void AddVectorObs(bool observation) { collectObservationsSensor.AddObservation(observation); } protected void AddVectorObs(int observation, int range) { collectObservationsSensor.AddOneHotObservation(observation, range); } /// /// Specifies the agent behavior at every step based on the provided /// action. /// /// /// Vector action. Note that for discrete actions, the provided array /// will be of length 1. /// public virtual void AgentAction(float[] vectorAction) { } /// /// Specifies the agent behavior when done and /// is false. This method can be /// used to remove the agent from the scene. /// public virtual void AgentOnDone() { } /// /// Specifies the agent behavior when being reset, which can be due to /// the agent or Academy being done (i.e. completion of local or global /// episode). /// public virtual void AgentReset() { } /// /// This method will forcefully reset the agent and will also reset the hasAlreadyReset flag. /// This way, even if the agent was already in the process of reseting, it will be reset again /// and will not send a Done flag at the next step. /// void ForceReset() { m_HasAlreadyReset = false; _AgentReset(); } /// /// An internal reset method that updates internal data structures in /// addition to calling . /// void _AgentReset() { ResetData(); m_StepCount = 0; AgentReset(); } public void UpdateAgentAction(AgentAction action) { m_Action = action; } /// /// Updates the vector action. /// /// Vector actions. public void UpdateVectorAction(float[] vectorActions) { m_Action.vectorActions = vectorActions; } /// /// Updates the value of the agent. /// public void UpdateValueAction(float value) { m_Action.value = value; } protected float GetValueEstimate() { return m_Action.value; } /// /// Scales continuous action from [-1, 1] to arbitrary range. /// /// /// /// /// protected float ScaleAction(float rawAction, float min, float max) { var middle = (min + max) / 2; var range = (max - min) / 2; return rawAction * range + middle; } /// /// Sets the status of the agent. Will request decisions or actions according /// to the Academy's stepcount. /// /// Number of current steps in episode void SetStatus(int academyStepCounter) { MakeRequests(academyStepCounter); } /// Signals the agent that it must reset if its done flag is set to true. void ResetIfDone() { // If an agent is done, then it will also // request for a decision and an action if (IsDone()) { if (agentParameters.resetOnDone) { if (agentParameters.onDemandDecision) { if (!m_HasAlreadyReset) { // If event based, the agent can reset as soon // as it is done _AgentReset(); m_HasAlreadyReset = true; } } else if (m_RequestDecision) { // If not event based, the agent must wait to request a // decision before resetting to keep multiple agents in sync. _AgentReset(); } } else { m_Terminate = true; RequestDecision(); } } } /// /// Signals the agent that it must sent its decision to the brain. /// void SendInfo() { if (m_RequestDecision) { SendInfoToBrain(); ResetReward(); m_Done = false; m_MaxStepReached = false; m_RequestDecision = false; m_HasAlreadyReset = false; } } /// Used by the brain to make the agent perform a step. void AgentStep() { if (m_Terminate) { m_Terminate = false; ResetReward(); m_Done = false; m_MaxStepReached = false; m_RequestDecision = false; m_RequestAction = false; m_HasAlreadyReset = false; OnDisable(); AgentOnDone(); } if ((m_RequestAction) && (m_Brain != null)) { m_RequestAction = false; AgentAction(m_Action.vectorActions); } if ((m_StepCount >= agentParameters.maxStep) && (agentParameters.maxStep > 0)) { m_MaxStepReached = true; Done(); } m_StepCount += 1; } /// /// Is called after every step, contains the logic to decide if the agent /// will request a decision at the next step. /// void MakeRequests(int academyStepCounter) { agentParameters.numberOfActionsBetweenDecisions = Mathf.Max(agentParameters.numberOfActionsBetweenDecisions, 1); if (!agentParameters.onDemandDecision) { RequestAction(); if (academyStepCounter % agentParameters.numberOfActionsBetweenDecisions == 0) { RequestDecision(); } } } void DecideAction() { m_Brain?.DecideAction(); } } }