Research map / Reinforcement learning / Visual and representation RL
APPLE: Toward General Active Perception via Reinforcement Learning
Tim Schneider, Cristiana de Farias, Roberto Calandra, Liming Chen, Jan Peters
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 |
|---|---|---|---|
| learn exploration with APPLE (SAC or CrossQ) instead of a random policy (ablation) | APPLE · CircleSquare | final accuracy (%): 68% -> 97% (SAC) | Improved |
| learn exploration with APPLE instead of a random policy (ablation) | APPLE · TactileMNIST | final accuracy (%): 74% -> 89% (CrossQ) | Improved |
| learn exploration with APPLE instead of a random policy (ablation) | APPLE · TactileMNISTVolume | final error (cm^3): 1.07 -> 0.99 (SAC); 1.05 (CrossQ) | No clear effect |
| use APPLE-CrossQ rather than SAC without tuning on the task (ablation) | APPLE · Toolbox | pose error | Improved |
| use CrossQ (no target networks) instead of SAC (ablation) | APPLE · training cost | training time: 53% reduction in training time on average | Improved |
| use on-policy PPO instead of off-policy SAC/CrossQ (ablation) | APPLE · CircleSquare | final accuracy (%): 79% (PPO) vs 96-97% (CrossQ, SAC) | Hurt |
| replace transformer with LSTM (HAM-style model) (ablation) | APPLE · CircleSquare | convergence speed: PPO: 79% -> 72% | Hurt |
| remove supervised prediction loss and label (pure RL) (ablation) | APPLE · CircleSquare | final accuracy (%): peaks at 80% | Hurt |
| replace learned exploration with static grid search (ablation) | APPLE · CircleSquare | final accuracy (%): stagnates at 73% | Hurt |
| share embeddings across Q-networks, action policy and prediction policy (ablation) | APPLE · tactile tasks | performance | Improved |
| use HAM (REINFORCE, LSTM) instead of APPLE | HAM · CircleSquare | final accuracy | Hurt |
Closest papers by meaning
- ACTIVE-o3 : Empowering MLLMs with Active Perception via Pure Reinforcement Learning (ICML 2026 · similarity 0.91)
- Real-World Reinforcement Learning of Active Perception Behaviors (NeurIPS 2025 · similarity 0.88)
- Enhancing Tactile-based Reinforcement Learning for Robotic Control (NeurIPS 2025 · similarity 0.88)
- Tactile MNIST: Benchmarking Active Tactile Perception (NeurIPS 2026 · similarity 0.87)
- Touch-R1: Reinforcing Touch Reasoning in MLLMs (NeurIPS 2026 · similarity 0.87)
- GUI-Eyes: Tool-Augmented Perception for Visual Grounding in GUI Agents (AAAI 2026 · 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.
