模倣するな、推定せよ：視覚運動制御のための微分可能な状態ベース方策の再利用
Estimate, Don't Imitate: Reusing Differentiable State-Based Policies for Visuomotor Control
シミュレーションで訓練した状態ベースの専門家方策を凍結して再利用し、その入力状態を画像から推定する視覚推定器を、状態教師信号と専門家を通した行動整合損失で訓練することで、接触の多い操作タスクの視覚運動制御を実現した。
詳しい要約
1. どんなもの？
2. 先行研究と比べてどこがすごい？
3. 技術・手法の肝は？
4. どうやって有効だと検証した？
5. 議論はある？
6. 次に読むべき論文は？
※ AIが要旨から生成した要約です。正確性は原文をご確認ください。
著者: Denis Shcherba, Adrian Abel, Eckart Cobo-Briesewitz, Wojciech Samek, Marc Toussaint
分類: cs.RO, cs.LG
原文アブストラクト
Simulation-trained manipulation policies can exploit privileged state information to learn effective contact-rich behaviours, but deployment requires acting from partial observations such as noisy camera images. A common solution is teacher-student distillation, in which a visuomotor policy is trained to reproduce the actions of the privileged expert. This requires the student to jointly infer the task-relevant state and relearn the expert's action mapping that is already available. An alternative is to reuse the state-based expert and learn only a perceptual interface that reconstructs its missing state inputs. However, minimising the state estimate error alone does not necessarily minimise the downstream control error induced by these estimates. To bridge this gap, we train a visual state estimator using both direct state supervision and an action-consistency loss backpropagated through the frozen, differentiable expert. A scheduled objective first establishes a physically meaningful state estimate and progressively emphasises errors that affect the expert's actions. Across five goal-conditioned manipulation tasks, retaining the expert consistently outperforms direct pixel-to-action imitation from the same expert demonstration corpus. We further demonstrate sim-to-real transfer on a physical Panda robot, achieving 76% success without retraining the underlying expert.