# Roll-Drop: accounting for observation noise with a single parameter

arXiv: https://arxiv.org/abs/2304.13150 · HTML: https://mlsanity.com/paper/2304.13150/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Builds on 21 of its 27 references that are in the map.
- First posted on arXiv 25 Apr 2023.

## Builds on (most cited first, 21 of 21)

- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning](https://mlsanity.com/paper/1506.02142/) (2015)
- [Domain Randomization for Transferring Deep Neural Networks from Simulation to the Real World](https://mlsanity.com/paper/1703.06907/) (2017)
- [Isaac Gym: High Performance GPU-Based Physics Simulation For Robot Learning](https://mlsanity.com/paper/2108.10470/) (2021)
- [Learning Quadrupedal Locomotion over Challenging Terrain](https://mlsanity.com/paper/2010.11251/) (2020)
- [Learning to Walk in Minutes Using Massively Parallel Deep Reinforcement Learning](https://mlsanity.com/paper/2109.11978/) (2021)
- [RMA: Rapid Motor Adaptation for Legged Robots](https://mlsanity.com/paper/2107.04034/) (2021)
- [Learning agile and dynamic motor skills for legged robots](https://mlsanity.com/paper/1901.08652/) (2019)
- [Learning robust perceptive locomotion for quadrupedal robots in the wild](https://mlsanity.com/paper/2201.08117/) (2022)
- [Sim-to-Real: Learning Agile Locomotion For Quadruped Robots](https://mlsanity.com/paper/1804.10332/) (2018)
- [Sim-to-Real Learning of All Common Bipedal Gaits via Periodic Reward Composition](https://mlsanity.com/paper/2011.01387/) (2020)
- [Concurrent Training of a Control Policy and a State Estimator for Dynamic and Robust Legged Locomotion](https://mlsanity.com/paper/2202.05481/) (2022)
- [Multi-expert learning of adaptive legged locomotion](https://mlsanity.com/paper/2012.05810/) (2020)
- [RLOC: Terrain-Aware Legged Locomotion using Reinforcement Learning and Optimal Control](https://mlsanity.com/paper/2012.03094/) (2020)
- [Learning Locomotion Skills Using DeepRL: Does the Choice of Action Space Matter?](https://mlsanity.com/paper/1611.01055/) (2016)
- [Robust Recovery Controller for a Quadrupedal Robot using Deep Reinforcement Learning](https://mlsanity.com/paper/1901.07517/) (2019)
- [Imitate and Repurpose: Learning Reusable Robot Movement Skills From Human and Animal Behaviors](https://mlsanity.com/paper/2203.17138/) (2022)
- [Crossing The Gap: A Deep Dive into Zero-Shot Sim-to-Real Transfer for Dynamics](https://mlsanity.com/paper/2008.06686/) (2020)
- [Learning and Deploying Robust Locomotion Policies with Minimal Dynamics Randomization](https://mlsanity.com/paper/2209.12878/) (2022)
- [An Efficient Locally Reactive Controller for Safe Navigation in Visual Teach and Repeat Missions](https://mlsanity.com/paper/2201.03938/) (2022)
- [Dynamic Bipedal Maneuvers through Sim-to-Real Reinforcement Learning](https://mlsanity.com/paper/2207.07835/) (2022)

## Built on by (most cited first, 3 of 3)

- [DreamWaQ++: Obstacle-Aware Quadrupedal Locomotion With Resilient Multi-Modal Reinforcement Learning](https://mlsanity.com/paper/2409.19709/) (2024)
- [Learning and Calibrating Heterogeneous Bounded Rational Market Behaviour with Multi-Agent Reinforcement Learning](https://mlsanity.com/paper/2402.00787/) (2024)
- [Acrobotics: A Generalist Approach to Quadrupedal Robots' Parkour](https://mlsanity.com/paper/2509.02727/) (2025)

## Abstract (arXiv)

This paper proposes a simple strategy for sim-to-real in Deep-Reinforcement Learning (DRL) -- called Roll-Drop -- that uses dropout during simulation to account for observation noise during deployment without explicitly modelling its distribution for each state. DRL is a promising approach to control robots for highly dynamic and feedback-based manoeuvres, and accurate simulators are crucial to providing cheap and abundant data to learn the desired behaviour. Nevertheless, the simulated data are noiseless and generally show a distributional shift that challenges the deployment on real machines where sensor readings are affected by noise. The standard solution is modelling the latter and injecting it during training; while this requires a thorough system identification, Roll-Drop enhances the robustness to sensor noise by tuning only a single parameter. We demonstrate an 80% success rate when up to 25% noise is injected in the observations, with twice higher robustness than the baselines. We deploy the controller trained in simulation on a Unitree A1 platform and assess this improved robustness on the physical system.