continuous-control

This paper introduces V-Simba, a visual RL architecture that adds normalization layers and pointwise convolutions to improve sample efficiency and stability. It matches or outperforms state-of-the-art methods across DMC, Adroit, and Meta-World benchmarks while being more computationally efficient than DrQ-v2.

This paper proposes OG-SPR, a model-free visual RL algorithm that combines latent self-prediction with observation prediction to learn dynamics-aware representations, achieving improved sample efficiency on DeepMind Control Suite tasks.

This paper proposes FaStR, a method that factorizes the transition kernel in reinforcement learning using CP decomposition into separate state, action, and next-state encoders, improving sample efficiency especially in high-dimensional locomotion tasks.

SteinGate introduces a distributional safety certificate using Kernelized Stein Discrepancy to detect rare catastrophic tail events in safe reinforcement learning, dynamically adapting policy updates to reduce constraint violations while maintaining competitive returns.

Delta-JEPA introduces a reconstruction-free world model that augments latent forward prediction with a Latent Difference Action Decoder to prevent collapse and improve action-sensitivity, achieving better planning performance on visual continuous-control tasks.

This paper proposes AutoSafe, a safety-aware policy architecture for safe online reinforcement learning that integrates structured safety monitoring and intervention directly into action generation, enabling smooth, risk-dependent transitions between performance and safety behaviors, demonstrated on benchmarks and a physical cart-pole system.

This paper presents low-power analogue neural networks that place trainable nonlinear functions on connections, inspired by Kolmogorov-Arnold networks, enabling efficient continuous control tasks with far fewer nodes and connections than multilayer perceptrons, demonstrated on hardware with projected microWatt power.

QGF is an RL algorithm that improves policies at test time by using a value gradient to guide a pre-trained flow policy, avoiding training-time instability while maintaining competitive performance.

GenPO++ proposes a reversible generative policy optimization framework that uses history states as auxiliary memory in a high-order reversible ODE solver, enabling exact inversion and Jacobian-free likelihood-ratio computation for flow-based policies in reinforcement learning. It achieves competitive performance on large-scale control, fine-tuning, and real-world robotic tasks while improving stability and efficiency.

This paper argues that representation learning, not model-based planning, is the key to scalable multitask deep reinforcement learning. It introduces MR.Q, a simple model-free algorithm with auxiliary predictive objectives that outperforms prior world-model-based methods across diverse continuous control tasks.

Introduced NOML, a custom reinforcement learning algorithm for continuous flight control that uses a hierarchical actor, anchor policy, and mirror learning to prevent oscillation and improve stability. The code is open-sourced on GitHub.

Introduces QuantFPFlow, a reinforcement learning framework that uses quantum amplitude estimation to achieve a quadratic speedup in estimating the Fokker-Planck partition function for continuous control, improving exploration and avoiding local optima.

Proposes R2R2, a regularization method for self-predictive learning in reinforcement learning to mitigate overfitting under high update-to-data ratios, achieving significant improvements on continuous control tasks.

OpenAI introduces a suite of challenging multi-goal reinforcement learning tasks for robotics using Fetch and Shadow Dexterous Hand hardware, integrated with OpenAI Gym, along with research directions for improving RL algorithms.