Inverse Dynamics Model
A model type opposed to the forward dynamics model, which solves inversely for the action to take given a current state and a target state. In hierarchical planning it serves as the low-level controller or solver, converting high-level subgoals into concrete actions.
Timeline (last 90 days)
The inverse dynamics model answers "what action should be taken from the current state to the goal state," in contrast to the forward dynamics model; in hierarchical planning it serves as the low-level controller that translates high-level subgoals into concrete actions.
An inverse dynamics model (IDM) is a class of models that infer the executed action from the observations or latent states at two adjacent time steps.
Action predictability training uses the inverse dynamics prediction error as an auxiliary loss, forcing the encoder to preserve action-discriminative state information.
All Facts (3)
The inverse dynamics model answers "what action should be taken from the current state to the goal state," in contrast to the forward dynamics model; in hierarchical planning it serves as the low-level controller that translates high-level subgoals into concrete actions.
50%UnverifiedAn inverse dynamics model (IDM) is a class of models that infer the executed action from the observations or latent states at two adjacent time steps.
50%UnverifiedAction predictability training uses the inverse dynamics prediction error as an auxiliary loss, forcing the encoder to preserve action-discriminative state information.
50%