Proposes a VLA fine-tuning technique that preserves pretrained VLM representations while aligning language to actions, raising real-robot generalization success from 28% to 54%.

Insect locomotion exhibits remarkable adaptability and flexibility despite the limited scale of its nervous system. However, the underlying principles that govern leg coordination remain difficult to extract and model. This paper presents an adversarial inverse reinforcement learning approach to infer transferable locomotor principles from limited insect behavioral data, enabling robust robot locomotion.

The UnderwaterVLA dual-brain vision-language-action architecture enables robust autonomous underwater navigation

Real-world robotic manipulation in homes and factories demands reliability, efficiency, and robustness that approach or surpass skilled human operators. We present a real-world reinforcement learning (RL) framework, RL-100, for achieving performant robotic manipulation through real-world RL training.

Coordinating multiple unmanned aerial vehicles (UAVs) for cooperative missions requires agents that perceive their environment, reason about objectives, and generate joint actions. Vision–language–action (VLA) models unify these capabilities but lack efficiency for real-world multi-UAV deployment. This episode explores MiniUAV-VLA, a compact 158M-parameter model that uses MARL expert distillation to enable cooperative multi-UAV search and elimination missions.

A neural dynamics world model paired with model-free RL policies for quadruped and humanoid locomotion in IsaacLab, enabling long-horizon autoregressive prediction and imagined rollouts that outperform pure model-based RL in prediction accuracy, policy learning, and sim-to-real transfer.

A 3B parameter unified decoder-only transformer that interleaves vision, text, and action tokens for perception, planning, reasoning, and control in a single model. Trained on the 1.5M-sample EO-1.5M dataset with strong results across manipulation tasks and benchmarks.

Introduces Video-Action Models (VAMs) that leverage pretrained internet-scale video models such as Cosmos-Predict2 as backbones instead of VLMs, paired with a flow-matching action decoder. Claims approximately 10x sample efficiency gains over standard vision-language-action models on real-world pick-and-place tasks.

Introduces FastDSAC, a method that scales maximum entropy reinforcement learning to high-dimensional humanoid control tasks, improving sample efficiency and policy performance.

Proposes a minimalist pipeline using retargeting-guided RL for dexterous manipulation in humanoid robots. Focuses on simplifying the training recipe while maintaining strong manipulation performance.