FImitationAgentState

Struct representing imitation state received from external source.

Contains observations, actions, rewards, and completion flags for imitation learning. This represents the full state for a single agent during one step of expert demonstration.

struct FImitationAgentState

Methods

Empty

inline void Empty()

Reset all fields to their default values.

Attributes

Observations

TInstancedStruct<FPoint> Observations

Observations received by the agent in the current state.

---

Actions

TInstancedStruct<FPoint> Actions

Expert actions provided for the agent to follow.

---

Reward

float Reward = 0.0f

Reward signal received for the current state.

---

bTerminated

bool bTerminated = false

Flag indicating whether the episode has completed normally.

---

bTruncated

bool bTruncated = false

Flag indicating whether the episode was truncated (e.g., timeout).

---

Info

TMap<FString, FString> Info

Additional metadata as key-value pairs.

Source: Source/ScholaImitation/Public/ImitationDataTypes/ImitationAgentState.h