Research map / Reinforcement learning / Visual and representation RL
Generation of Geodesics with Actor-Critic Reinforcement Learning to Predict Midpoints
Kazumi Kasaura
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 cycle-based instead of timestep-based depth scheduling (ablation) | midpoint tree actor-critic · Matsumoto | pairwise percentage of shorter paths: 37 ± 3 | Improved |
| use cycle-based instead of timestep-based depth scheduling (ablation) | midpoint tree actor-critic · Car-Like | pairwise percentage of shorter paths: 18 ± 3 | Improved |
| predict midpoints (squared loss) instead of arbitrary intermediate points (ablation) | midpoint tree actor-critic · 2D Obstacles | pairwise percentage of shorter paths: 42 ± 15 | Improved |
| predict midpoints instead of 2:1 division points (ablation) | midpoint tree actor-critic · Matsumoto | pairwise percentage of shorter paths: 58 ± 5 | Improved |
| add midpoint-tree actor-critic over sequential PPO | Our-C · Matsumoto | pairwise percentage of shorter paths: 73 ± 3 | Improved |
| use sequential PPO instead of Our-T | Seq · 2D Obstacles | pairwise percentage of shorter paths: 32 ± 10 | Improved |
| add a critic to policy-gradient midpoint prediction (ablation) | midpoint tree · all environments | pairwise percentage of shorter paths | Improved |
| add cut cost c_cut to intermediate point prediction (ablation) | Inter · car-like and higher-dimensional environments | pairwise percentage of shorter paths | No clear effect |
Closest papers by meaning
- Adversarial Actor-Critic Method for Task and Motion Planning Problems Using Planning Experience (AAAI 2019 · similarity 0.86)
- Geometry of Neural Reinforcement Learning in Continuous State and Action Spaces (ICLR 2025 · similarity 0.84)
- MAPF-GPT: Imitation Learning for Multi-Agent Pathfinding at Scale (AAAI 2025 · similarity 0.83)
- Actor-Critics Can Achieve Optimal Sample Efficiency (ICML 2025 · similarity 0.83)
- Massively Scalable Inverse Reinforcement Learning in Google Maps (ICLR 2024 · similarity 0.83)
- Learn to Follow: Decentralized Lifelong Multi-Agent Pathfinding via Planning and Learning (AAAI 2024 · similarity 0.83)
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.
