experiments/METHOD_AUDIT.md

FlowMo Method And Baseline Audit

This document records what each paper-facing method implements and what comparison it supports.

A. Learned World Models

Method	Implemented Architecture	Rollout State	Transition Form	Comparison Purpose
FlowMo	Shared image encoder; short state GRU; long strided context GRU; base transition plus zero-context residual	z_t plus c_t	z + F0(z,a) + R(z,c) - R(z,0)	Tests whether separating short object motion from long ambient drift improves prediction and planning.
LeWorldModel	Image encoder; current-frame latent state; action-conditioned residual transition	z_t	z + F(z,a)	Tests a simple JEPA-style image-latent predictor without temporal drift context.
PlaNet RSSM	Image encoder; deterministic recurrent state; stochastic latent posterior/prior; decoder from recurrent plus stochastic state	deterministic recurrent state plus stochastic latent	recurrent transition plus prior/posterior latent	Tests generic recurrent state-space memory under the shared rollout protocol.
TD-MPC2 Dynamics	Image encoder; short action-conditioned history encoder; residual latent dynamics	z_t	z + F(z,a) after history encoding	Tests compact action-conditioned latent dynamics without explicit drift context.

B. Traditional Non-WM Controllers

Method	Input	Controller Type	Comparison Purpose
PID/LOS controller	Clean image pose estimate	Line-of-sight waypoint tracking	Simple hand-designed tracking baseline.
No-Flow LOS Controller	Clean image pose estimate	Line-of-sight tracking without flow compensation	Measures the cost of ignoring external current.
Current-Estimator LOS Controller	Clean image pose estimate and recent drift	Line-of-sight tracking with online drift estimate	Strong classical current-compensation baseline.
Oracle-Flow LOS Controller	Clean image pose estimate and simulator local flow	Line-of-sight tracking with true local flow feed-forward	Privileged-information reference for how much true local flow helps a simple geometric controller.

Architecture Difference Checklist

Method	Uses History	Has Long Context	Has Explicit Drift Context	Distinct From FlowMo
FlowMo	Yes, short state history	Yes, strided long context	Yes, c_t	Proposed factorization.
LeWorldModel	No, current image only	No	No	Tests no-history latent prediction.
PlaNet RSSM	Yes	No	No	Tests generic recurrent memory.
TD-MPC2 Dynamics	Yes, short history	No	No	Tests compact latent dynamics.