Survival Reinforcement Learning: Toward Scalable Self-Supervised RL
Abstract
While self-supervised Contrastive Reinforcement Learning (CRL) has shown remarkable depth-scaling capabilities, successfully using networks over 64 layers, scaled CRL still struggles with long-horizon goal-conditioned planning due to the uniformity-tolerance dilemma inherent in contrastive losses. We introduce Survival Reinforcement Learning (SRL), an online classification-based alternative that extends the survival value learning framework by maximizing the agent's dwell time at target goals. SRL bypasses the structural constraints of CRL and mitigates the "bang-bang" control solutions inherent to survival frameworks, which often induce undesirable behavior in complex dynamical systems. Evaluated across diverse robotic benchmarks, scaled SRL matches state-of-the-art CRL on manipulation tasks and outperforms it by 2x to 8x on stable, long-horizon locomotion tasks. Our results provide strong additional evidence that classification-based methods may serve as a key primitive in the broader effort to scale reinforcement learning and an open-source implementation is available online: https://github.com/Simple-Robotics/survival-reinforcement-learning.
Related
- Scalable Self-Supervised Learning for Multiphase AC-OPF in Distribution Systems with Topology Reconfiguration
- Scalable Multi-Task Inverse Reinforcement Learning
- Learning a Flow to Self-Supervised Representations
- Is Spectral Self-Supervised Learning a Minimax Optimal Method of Density Ratio Estimation?
- Evaluating Xception with Multitask Self-Supervised Learning for AI-Generated Image and Deepfake Video Detection
BibTeX
@misc{nguimatsiatiofack2026survival,
title = {{Survival Reinforcement Learning: Toward Scalable Self-Supervised RL}},
author = {Franki Nguimatsia-Tiofack and Fabian Schramm and Théotime Le Hellard and Justin Carpentier},
howpublished = {HAL},
year = {2026},
month = sep,
eprint = {2605.31273},
archivePrefix = {arXiv},
url = {https://hal.science/hal-05649866v2},
note = {CC BY-NC-SA 4.0},
}
RIS
TY - UNPB
TI - Survival Reinforcement Learning: Toward Scalable Self-Supervised RL
AU - Franki Nguimatsia-Tiofack
AU - Fabian Schramm
AU - Théotime Le Hellard
AU - Justin Carpentier
PY - 2026
DA - 2026/09/18
PB - HAL
UR - https://hal.science/hal-05649866v2
AB - While self-supervised Contrastive Reinforcement Learning (CRL) has shown remarkable depth-scaling capabilities, successfully using networks over 64 layers, scaled CRL still struggles with long-horizon goal-conditioned planning due to the uniformity-tolerance dilemma inherent in contrastive losses. We introduce Survival Reinforcement Learning (SRL), an online classification-based alternative that extends the survival value learning framework by maximizing the agent's dwell time at target goals. SRL bypasses the structural constraints of CRL and mitigates the "bang-bang" control solutions inherent to survival frameworks, which often induce undesirable behavior in complex dynamical systems. Evaluated across diverse robotic benchmarks, scaled SRL matches state-of-the-art CRL on manipulation tasks and outperforms it by 2x to 8x on stable, long-horizon locomotion tasks. Our results provide strong additional evidence that classification-based methods may serve as a key primitive in the broader effort to scale reinforcement learning and an open-source implementation is available online: https://github.com/Simple-Robotics/survival-reinforcement-learning.
ER -