Research map / Reinforcement learning / Visual and representation RL
Progressive Reinforcement Learning with Distillation for Multi-Skilled Motion Control
Glen Berseth, Cheng Xie, Paul Cernek, Michiel Van de Panne
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 changed | Where | Measured | Effect |
|---|---|---|---|
| use PLAID's sequential transfer-learning and repeated policy-distillation schedule instead of TLOnly | motion control policy · Average over terrains | average reward: 0.64731222 -> 0.6975943526 | Improved |
| use PLAID's sequential transfer-learning and repeated policy-distillation schedule instead of MultiTasker | motion control policy · Average over terrains | average reward: 0.6876953277 -> 0.6975943526 | Improved |
| use PLAID's progressive schedule instead of TLOnly with a single final distillation | motion control policy · Average over terrains | average reward: 0.7234199776 -> 0.6975943526 | Hurt |
| use PLAID instead of TLOnly | motion control policy · Gaps | average reward: 0.62609913758 -> 0.5289199521 | Hurt |
| use PLAID instead of MultiTasker | motion control policy · Slopes | average reward: 0.65631263654 -> 0.60244157756 | Hurt |
| add a final distillation step after sequential transfer learning (ablation) | TLOnly · Average over terrains | average reward: 0.64731222 -> 0.7234199776 | Improved |
| reuse value functions when transferring to a new task | PLAID transfer learning · new locomotion task | learning performance | No clear effect |
Closest papers by meaning
- Hierarchical Skills for Efficient Exploration (NeurIPS 2021 · similarity 0.88)
- Reward-Adaptive Reinforcement Learning: Dynamic Policy Gradient Optimization for Bipedal Locomotion (TPAMI 2022 · similarity 0.88)
- Knowledge Transfer for Deep Reinforcement Learning with Hierarchical Experience Replay (AAAI 2017 · similarity 0.88)
- Training Transition Policies via Distribution Matching for Complex Tasks (ICLR 2022 · similarity 0.88)
- Adversarial Dual On-Policy Distillation from Expressive Teacher (ICML 2026 · similarity 0.88)
- Acquiring Diverse Skills using Curriculum Reinforcement Learning with Mixture of Experts (ICML 2024 · similarity 0.86)
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.
