Supplementary file 1_Morphological symmetry-aware generalized policy network for deep rein

A PDF document authored by Ryo Hakoda, published on figshare under a CC-BY-4.0 license on May 13, 2026. The file describes a symmetry-assisted deep reinforcement learning framework designed for stable and robust control of morphologically symmetric robots, such as humanoids and quadrupeds. It proposes modeling the environment as a symmetric Markov decision process and constructing a full-body policy from a single-sided base policy.

Use Cases

• Implementing symmetry-aware policy networks for robot control based on the described framework.
• Benchmarking new DRL algorithms against the proposed symmetric PPO objective.
• Applying symmetry operators to generalize policies for humanoid or quadruped robots.
• Studying the stability of DRL training in symmetric Markov decision processes.

Strengths

• File size is 295.1 KB, indicating a concise document.
• License is CC-BY-4.0, permitting open sharing and adaptation.
• Description provides a detailed methodological proposal for a specific robotics problem.

Limitations

• The dataset is a single PDF file; no structured data tables are provided.
• Row count and column-level documentation are absent, limiting direct data analysis.
• Freshness should be verified; last updated 2026-05-13 07:53:15.

Provenance

Source

figshare

Freshness

Last updated 2026-05-13 07:53:15.