# Cognitively-Grounded On-Device Runtime Learning for Ground Robots in Unknown Physical Environments

arXiv: https://arxiv.org/abs/2608.28677 · HTML: https://mlsanity.com/paper/2608.28677/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Safe and constrained RL, one of 15 directions in Deep reinforcement learning methods.
- Builds on 22 of its 47 references that are in the map.
- First posted on arXiv 25 Aug 2026.

## Placed in

- Deep reinforcement learning methods / [Safe and constrained RL](https://mlsanity.com/deeprl/safe-and-constrained-rl/)

## Builds on (most cited first, 22 of 22)

- [Sim-to-Real: Learning Agile Locomotion For Quadruped Robots](https://mlsanity.com/paper/1804.10332/) (2018)
- [Learning to Adapt in Dynamic, Real-World Environments Through Meta-Reinforcement Learning](https://mlsanity.com/paper/1803.11347/) (2018)
- [Real-World Humanoid Locomotion with Reinforcement Learning](https://mlsanity.com/paper/2303.03381/) (2023)
- [Rapid Locomotion via Reinforcement Learning](https://mlsanity.com/paper/2205.02824/) (2022)
- [ANYmal Parkour: Learning Agile Navigation for Quadrupedal Robots](https://mlsanity.com/paper/2306.14874/) (2023)
- [Lifelong Federated Reinforcement Learning: A Learning Architecture for Navigation in Cloud Robotic Systems](https://mlsanity.com/paper/1901.06455/) (2019)
- [Remember and Forget for Experience Replay](https://mlsanity.com/paper/1807.05827/) (2018)
- [Control Regularization for Reduced Variance Reinforcement Learning](https://mlsanity.com/paper/1905.05380/) (2019)
- [Safe Reinforcement Learning for Legged Locomotion](https://mlsanity.com/paper/2203.02638/) (2022)
- [Neural Simplex Architecture](https://mlsanity.com/paper/1908.00528/) (2019)
- [Vision-Guided Quadrupedal Locomotion in the Wild with Multi-Modal Delay Randomization](https://mlsanity.com/paper/2109.14549/) (2021)
- [Bayesian Controller Fusion: Leveraging Control Priors in Deep Reinforcement Learning for Robotics](https://mlsanity.com/paper/2107.09822/) (2021)
- [Dynamic Model Predictive Shielding for Provably Safe Reinforcement Learning](https://mlsanity.com/paper/2405.13863/) (2024)
- [Physics-Regulated Deep Reinforcement Learning: Invariant Embeddings](https://mlsanity.com/paper/2305.16614/) (2023)
- [SL1-Simplex: Safe Velocity Regulation of Self-Driving Vehicles in Dynamic and Unforeseen Environments](https://mlsanity.com/paper/2008.01627/) (2020)
- [Runtime Safety Assurance for Learning-enabled Control of Autonomous Driving Vehicles](https://mlsanity.com/paper/2109.13446/) (2021)
- [Equipping Black-Box Policies with Model-Based Advice for Stable Nonlinear Control](https://mlsanity.com/paper/2206.01341/) (2022)
- [Cloud-Edge Training Architecture for Sim-to-Real Deep Reinforcement Learning](https://mlsanity.com/paper/2203.02230/) (2022)
- [Lane-Keeping Control of Autonomous Vehicles Through a Soft-Constrained Iterative LQR](https://mlsanity.com/paper/2311.16900/) (2023)
- [Interactive-FAR:Interactive, Fast and Adaptable Routing for Navigation Among Movable Obstacles in Complex Unknown Environments](https://mlsanity.com/paper/2404.07447/) (2024)
- [Sample Efficient Experience Replay in Non-stationary Environments](https://mlsanity.com/paper/2509.15032/) (2025)
- [Demonstrating Reinforcement Learning and Run Time Assurance for Spacecraft Inspection Using Unmanned Aerial Vehicles](https://mlsanity.com/paper/2405.06770/) (2024)

## Abstract (arXiv)

This paper presents ulCogRun, a framework that enables safety-critical ground robots to perform cognitively-grounded runtime learning entirely on edge-AI devices in unknown physical environments, without prior maps or perceptual knowledge. CogRun consists of three components: a Learning-Agent, a Rational-Agent, and a Coordinator. The Learning-Agent is novel in cognitive-neural learning architecture, which featurs dedicated replay buffers, cognition-driven experience sampling, and a safety-aware action blending of actor-critic reinforcement learning (RL) with instance-based learning (IBL). The Rational-Agent is a non-learning module that complements the Learning-Agent by exclusively handling safety-critical functions, while the Coordinator manages interactions between the two agents to promote safe and efficient runtime learning. CogRun's full autonomy stack (i.e., perception, learning, and control) on edge-AI devices eliminates dependence on wireless communications, enabling broader applications in challenging environments with limited or no connectivity. Experiments on a quadruped robot in real-world wild forests and on an off-road autonomous vehicle in a simulated wild forest demonstrate that CogRun enables safe and efficient runtime learning, allowing robots to safely and continuously interact with the physical world for enhancing task performance in complex, unknown environments.