AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

Neural Dynamic Policies for End-to-End Sensorimotor Learning

Shikhar Bahl, Mustafa Mukadam, Abhinav Gupta, Deepak Pathak

NeurIPS 2020 · Reinforcement learning › Visual and representation RL · approach: Vision-based robot RL · 9 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Use NDP instead of NNbehavior cloning · Throwingsuccess rate: 0.528 $\pm$ 0.262 -> 0.642 $\pm$ 0.246Improved
Use NDP instead of NNbehavior cloning · Pickingsuccess rate: 0.672 $\pm$ 0.074 -> 0.408 $\pm$ 0.058Hurt
Use NDP instead of NNbehavior cloning · Pushingsuccess rate: 0.002 $\pm$ 0.004 -> 0.208 $\pm$ 0.049Improved
Use NDP instead of NNbehavior cloning · Soccersuccess rate: 0.885 $\pm$ 0.016 -> 0.890 $\pm$ 0.010No clear effect
Use NDP instead of NNbehavior cloning · Faucet-Openingsuccess rate: 0.532 $\pm$ 0.231 -> 0.790 $\pm$ 0.059Improved
Predict dynamical-system parameters with NDP rather than a CNNdigit-writing behavior cloning · Digit writing trainmean trajectory loss: 10.42 $\pm$ 5.26 -> 0.70 $\pm$ 0.36Improved
Predict dynamical-system parameters with NDP rather than CNN-DMPdigit-writing behavior cloning · Digit writing held-out testmean trajectory loss: 8.46 $\pm$ 8.45 -> 0.74 $\pm$ 0.34Improved
Remove the forcing term and learn only the goal g (ablation)NDP · Pushingsample efficiencyHurt
Remove the forcing term and learn only the goal g (ablation)NDP · Pushingasymptotic performanceHurt
Learn global α instead of keeping it fixed (ablation)NDP · PushingperformanceHurt
Reduce rollout length to k=3 (ablation)NDP · PushingperformanceHurt
Vary the number of basis functions, integration steps, and basis-function type (ablation)NDP · PushingperformanceNo clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy