# E2DRL: End-to-End Differentiable Reinforcement Learning for 3D Obstacle Avoidance in Quadruped Robots

arXiv: https://arxiv.org/abs/2610.19627 · HTML: https://mlsanity.com/paper/2610.19627/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in RL for robot navigation and UAV control, one of 15 directions in Deep reinforcement learning methods.
- Builds on 23 of its 30 references that are in the map.
- First posted on arXiv 14 Oct 2026.

## Placed in

- Deep reinforcement learning methods / [RL for robot navigation and UAV control](https://mlsanity.com/deeprl/rl-for-robot-navigation-and-uav-control/)

## Builds on (most cited first, 23 of 23)

- [Soft Actor-Critic Algorithms and Applications](https://mlsanity.com/paper/1812.05905/) (2018)
- [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)
- [Brax -- A Differentiable Physics Engine for Large Scale Rigid Body Simulation](https://mlsanity.com/paper/2106.13281/) (2021)
- [Learning robust perceptive locomotion for quadrupedal robots in the wild](https://mlsanity.com/paper/2201.08117/) (2022)
- [Extreme Parkour with Legged Robots](https://mlsanity.com/paper/2309.14341/) (2023)
- [Robot Parkour Learning](https://mlsanity.com/paper/2309.05665/) (2023)
- [Walk These Ways: Tuning Robot Control for Generalization with Multiplicity of Behavior](https://mlsanity.com/paper/2212.03238/) (2022)
- [Accelerated Policy Learning with Parallel Differentiable Simulation](https://mlsanity.com/paper/2204.07137/) (2022)
- [Do Differentiable Simulators Give Better Policy Gradients?](https://mlsanity.com/paper/2202.00817/) (2022)
- [Agile But Safe: Learning Collision-Free High-Speed Legged Locomotion](https://mlsanity.com/paper/2401.17583/) (2024)
- [Back to Newton's Laws: Learning Vision-based Agile Flight via Differentiable Physics](https://mlsanity.com/paper/2407.10648/) (2024)
- [Learning Quadruped Locomotion Using Differentiable Simulation](https://mlsanity.com/paper/2403.14864/) (2024)
- [Dojo: A Differentiable Physics Engine for Robotics](https://mlsanity.com/paper/2203.00806/) (2022)
- [Learning to walk in confined spaces using 3D representation](https://mlsanity.com/paper/2403.00187/) (2024)
- [Stabilizing Reinforcement Learning in Differentiable Multiphysics Simulation](https://mlsanity.com/paper/2412.12089/) (2024)
- [An Efficient Locally Reactive Controller for Safe Navigation in Visual Teach and Repeat Missions](https://mlsanity.com/paper/2201.03938/) (2022)
- [Residual Policy Learning for Perceptive Quadruped Control Using Differentiable Simulation](https://mlsanity.com/paper/2410.03076/) (2024)
- [Walking in Narrow Spaces: Safety-critical Locomotion Control for Quadrupedal Robots with Duality-based Optimization](https://mlsanity.com/paper/2212.14199/) (2022)
- [Learning Deployable Locomotion Control via Differentiable Simulation](https://mlsanity.com/paper/2404.02887/) (2024)
- [From Distances to Trajectories: Real-Time Signed Distance Function Mapping and Distance-Accelerated Motion Planning for UAVs](https://mlsanity.com/paper/2607.19306/) (2026)

## Abstract (arXiv)

Differentiable reinforcement learning shows strong promise for robot policy learning, but its application to legged locomotion has remained limited to blind walking or simulated 2D tasks. Extending it to end-to-end visual obstacle avoidance is challenging because hard collision constraints and early termination upon contact corrupt first-order gradients, trapping optimizers in local minima. We present E2DRL, which addresses this with a soft differentiable obstacle-interaction formulation: penetration penalties across multiple body keypoints with velocity-conditioned stop-gradient shaping to provide the first-order training signals for 3D obstacle avoidance. To realize a deployable perception-control system, a privileged teacher trained via Soft Analytic Policy Optimization (SAPO) in a differentiable simulator is distilled into a recurrent depth-based student via DAgger, then blended with a base locomotion policy using a terrain-aware gating module for deployment. In simulation, E2DRL achieves higher sample efficiency than PPO and demonstrates superior velocity-tracking performance and 3D obstacle-avoidance capability compared with both PPO and Duality-based DCBF NMPC. We present the first sim-to-real validation of end-to-end differentiable RL for 3D obstacle avoidance on a Unitree Go2 in static and dynamic environments.