analiza aiqbench
This paper trains quadrotor control policies through a differentiable simulator, treating obstacle avoidance as a hard constraint solved with primal-dual optimization rather than a hand-tuned penalty. The notable technical contributions are a bounded clearance constraint and log-mean-exp aggregation, which together let practitioners set a target safety level directly. For ML readers, the main points are the constrained-optimization framing and the reported zero-shot transfer to a physical drone, though the evidence is limited to the abstract and the convergence guarantee holds only under stated assumptions.
Constrained Differentiable Policy Learning for Safe Quadrotor Navigation
arXiv:2610.16119v1 Announce Type: new Abstract: Differentiable simulation enables efficient policy optimization through analytic gradients, yet safety is commonly incorporated as a weighted penalty, requiring manual tuning of the trade-off between task performance and constraint satisfaction. We instead formulate obstacle avoidance as an explicit constraint and train quadrotor policies using primal--dual gradient descent--ascent through a differentiable simulator. We introduce two key components to enable effective constrained training. First, we propose a bounded clearance-based constraint function that provides informative gradients around the safety boundary while preventing large violations from destabilizing primal--dual optimization. Second, we introduce log-mean-exp (LME) aggregation to smoothly emphasize critical per-step safety violations across a trajectory and training batch within a single constraint governed by one dual variable. Together, these components enable the desired safety level to be specified directly while balancing constraint satisfaction and task completion. We theoretically prove that the proposed optimization with differentiable simulation converges to feasible policy under mild assumptions. Experiments demonstrate improved safety--performance trade-offs over fixed-penalty and existing policy-learning baselines, as well as generalization to denser out-of-distribution environments. Finally, the learned policy transfers zero-shot to Gazebo/PX4 and a physical quadrotor without additional fine-tuning.
Abstract:Differentiable simulation enables efficient policy optimization through analytic gradients, yet safety is commonly incorporated as a weighted penalty, requiring manual tuning of the trade-off between task performance and constraint satisfaction. We instead formulate obstacle avoidance as an explicit constraint and train quadrotor policies using primal--dual gradient descent--ascent through a differentiable simulator. We introduce two key components to enable effective constrained training. First, we propose a bounded clearance-based constraint function that provides informative gradients around the safety boundary while preventing large violations from destabilizing primal--dual optimization. Second, we introduce log-mean-exp (LME) aggregation to smoothly emphasize critical per-step safety violations across a trajectory and training batch within a single constraint governed by one dual variable. Together, these components enable the desired safety level to be specified directly while balancing constraint satisfaction and task completion. We theoretically prove that the proposed optimization with differentiable simulation converges to feasible policy under mild assumptions. Experiments demonstrate improved safety--performance trade-offs over fixed-penalty and existing policy-learning baselines, as well as generalization to denser out-of-distribution environments. Finally, the learned policy transfers zero-shot to Gazebo/PX4 and a physical quadrotor without additional fine-tuning.
Robotics (cs.RO)
arXiv:2610.16119 [cs.RO]
(or arXiv:2610.16119v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2610.16119 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 276 cuvinte
Subiecte conexe
ClearNav: Perception-Aligned Quadrotor Flight Using Differential-Physics
arXiv · 15 oct. 2026
SEA-Nav: Efficient Policy Learning for Safe and Agile Quadruped Navigation in Cluttered Environments
arXiv · 15 sept. 2026
SafeDiSTL: Decomposing Joint Constraints for Safe Multi-Agent Reinforcement Learning
arXiv · 12 oct. 2026
The J-Space Debate, Agent Swarms, and Pacing Frontier AI - Digital Minds Newsletter #4
LessWrong · 18 sept. 2026