# A Balanced Data Diet: Addressing the Exploration Bottleneck in Mega-Scale RL for Robot Control

arXiv: https://arxiv.org/abs/2610.12465 · HTML: https://mlsanity.com/paper/2610.12465/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 19 of its 52 references that are in the map.
- First posted on arXiv 8 Oct 2026.

## Builds on (most cited first, 19 of 19)

- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [Isaac Gym: High Performance GPU-Based Physics Simulation For Robot Learning](https://mlsanity.com/paper/2108.10470/) (2021)
- [Solving Rubik's Cube with a Robot Hand](https://mlsanity.com/paper/1910.07113/) (2019)
- [RMA: Rapid Motor Adaptation for Legged Robots](https://mlsanity.com/paper/2107.04034/) (2021)
- [Orbit: A Unified Simulation Framework for Interactive Robot Learning Environments](https://mlsanity.com/paper/2301.04195/) (2023)
- [ManiSkill3: GPU Parallelized Robotics Simulation and Rendering for Generalizable Embodied AI](https://mlsanity.com/paper/2410.00425/) (2024)
- [BeyondMimic: From Motion Tracking to Versatile Humanoid Control via Guided Diffusion](https://mlsanity.com/paper/2508.08241/) (2025)
- [Intrinsic Motivation and Automatic Curricula via Asymmetric Self-Play](https://mlsanity.com/paper/1703.05407/) (2017)
- [Learning Montezuma's Revenge from a Single Demonstration](https://mlsanity.com/paper/1812.03381/) (2018)
- [DextrAH-RGB: Visuomotor Policies to Grasp Anything with Dexterous Hands](https://mlsanity.com/paper/2412.01791/) (2024)
- [BFM-Zero: A Promptable Behavioral Foundation Model for Humanoid Control Using Unsupervised Reinforcement Learning](https://mlsanity.com/paper/2511.04131/) (2025)
- [DexPBT: Scaling up Dexterous Manipulation for Hand-Arm Systems with Population Based Training](https://mlsanity.com/paper/2305.12127/) (2023)
- [Parkour in the Wild: Learning a General and Extensible Agile Locomotion Policy Using Multi-expert Distillation and RL Fine-tuning](https://mlsanity.com/paper/2505.11164/) (2025)
- [VIRAL: Visual Sim-to-Real at Scale for Humanoid Loco-Manipulation](https://mlsanity.com/paper/2511.15200/) (2025)
- [Reverse Forward Curriculum Learning for Extreme Sample and Demonstration Efficiency in Reinforcement Learning](https://mlsanity.com/paper/2405.03379/) (2024)
- [Towards bridging the gap: Systematic sim-to-real transfer for diverse legged robots](https://mlsanity.com/paper/2509.06342/) (2025)
- [Emergent Dexterity via Diverse Resets and Large-Scale Reinforcement Learning](https://mlsanity.com/paper/2603.15789/) (2026)
- [Combined Constrained Sampling and Reinforcement Learning for Robotic Manipulation](https://mlsanity.com/paper/2602.08557/) (2026)
- [Preventing Learning Stagnation in PPO by Scaling to 1 Million Parallel Environments](https://mlsanity.com/paper/2603.06009/) (2026)

## Abstract (arXiv)

General-purpose robots must perform a wide range of tasks from agile locomotion to dexterous manipulation. While sim-to-real reinforcement learning (RL) has proven to be a useful tool for this goal, current RL pipelines depend on engineering-heavy, per-task structural priors such as shaped rewards and demonstrations. Recent work has shown that diverse simulator resets, combined with massively parallel simulation, can alleviate much of this engineering burden on several manipulation problems. However, we find that naively scaling this paradigm to more precise or dynamic problems remains non-trivial. While simulator resets can help with exploration, uniformly sampling over this distribution wastes a growing fraction of learning experience on task configurations the policy has already mastered or cannot yet attempt. This makes it challenging to see the expected benefits of scaling parallel environments for RL, since much of the learning signal in a batch is wasted during learning. To mitigate this, we introduce Success Guided Sampling (SGS), a simple adaptive sampler that concentrates RL training on task configurations around the frontier of the policy's capabilities. Doing so allows large-scale simulated RL to make the most out of the experience in a batch, enabling much more effective scaling to large-scale parallel simulation. Across experiments using up to 220 (over one million) parallel environments, SGS enables RL to solve challenging multi-terrain quadruped locomotion and contact-rich assembly tasks that prior methods fail to solve. Finally, we distill the learned manipulation policies into RGB-based policies and demonstrate zero-shot transfer to several challenging assembly tasks on real hardware. Project website: https://sgs-rl.github.io/.