OpenFlyV2 flies a quadrotor onboard from vision and language
Authors report 48% zero-shot navigation success in four real environments with no real aerial training data.
Stories per day · 30 days
2 RSS
Relatedrobotics7dexterous-manipulation3simulation3active-perception1benchmark1drones1egocentric-video1gpu1isaac-lab1language-conditioning1leap-hand1manipulation1motion-tracking1nvidia1onboard-inference1quadrotor1ray-tracing1reinforcement-learning1
20 stories, newest first · one card per story
Stories tagged #sim-to-real
LEAP hand policies reach in-hand poses and cut object drops
Over 4,800 unseen goals, a residual filter cuts drops 75% while still reaching 91.5% of goals.
ActiveAlign uses motion and touch to close Real2Sim2Real gaps
On a Unitree G1, the method supports button presses and slider moves at millimeter scale.
Microduck trains in simulation and deploys with Isaac Lab 3.0
No post body was available, so only the sim-to-real Microduck and Isaac Lab 3.0 topic is known.
AdaDex turns short human demos into adaptable dexterous skills
One policy trains on motion tracking and noised goal environments, then transfers to real robots.
HyperDrone claims 20x faster GPU quadrotor simulation and sim-to-real transfer
Authors say HyperLux still hits millions of frames per second on consumer GPUs at highest fidelity.
ScaleNav pretrains language-conditioned robot navigation on internet video
Authors report 86.5% success on adapted Short-Horizon OVON and 94.9% on HabitatGS-LN, then tested a real quadruped.
HandsOn benchmark measures cross-embodiment dexterous manipulation
The best trained multi-task policy reached 29.1% Core-50 success; untuned GPT-6 scored 5.2%.
A Balanced Data Diet: Addressing the Exploration Bottleneck in Mega-Scale RL for Robot Control
Success Guided Sampling focuses mega-scale robot RL on the policy frontier, solving tasks prior methods miss.
Dex-One2Many: Learning Dexterous Manipulation from a Single Human Demonstration
Dex-One2Many learns a generalizable dexterous manipulation policy from one human video via scene-graph-guided reinforcement learning.
Skill2Real: Agentic Skill Learning for Zero-Shot Sim-to-Real Robot Manipulation
Skill2Real learns sim-to-real robot manipulation skills via a Proposer-Verifier-Governor loop, achieving 78.75% task completion on real robots without fine-tuning.
Behavioral Persistence and Incomplete Functional Transfer of Co-evolved Communication in Evolutionary Robotics
Co-evolved robot communication transferred from 2D simulation to 3D hardware only partially, without weight retraining.
Counterfactual Video Generation Enables Scalable Humanoid Loco-Manipulation
PRISM turns a few real videos into counterfactual data so a humanoid carries novel objects without real-world fine-tuning.
X-Reset: Scaling Object-Centric Reinforcement Learning via Cross-Embodiment Resets
X-Reset uses retargeted human hand-object states as RL resets to train cross-embodiment dexterous manipulation policies.
Morphometric Imitation: From Morphology and Contact Aware Hand Retargeting to Sim-to-Real Visuomotor Policy
Morphometric Imitation turns human hand-object demos into robot policies with 89.3% zero-shot success in 300 real trials.
Visuomotor Robotic Pruning in Planar Orchards Using Hybrid Reinforcement Learning
A sim-trained hybrid RL policy prunes planar orchard trees and transfers zero-shot to real trials.
MILER: Semantic Mid-Level Representation for Sim-to-Real Reinforcement Learning in Unstructured Autonomous Driving
MILER achieves zero-shot sim-to-real reinforcement learning for autonomous driving, completing 17.3 km on a test track without human intervention.
Fingers as Legs: Learning Self-Supported Locomotion and Manipulation with an Anthropomorphic Hand
Researchers trained an anthropomorphic robotic hand via reinforcement learning to crawl, steer, and recover from falls using its fingers as legs.
TANGO: Humanoid Navigation in Cluttered Environments with a Whole-Body Vision-Language-Action Model
Researchers introduce TANGO, a whole-body vision-language-action model enabling zero-shot language-guided humanoid navigation on the Unitree G1 robot.
OpenWAM: An Open, Modular Exploration Towards Systematic World-Action Model Pretraining
OpenWAM releases an open modular stack for world-action model pretraining, plus OpenWAM-alpha trained on about 6,400 hours of egocentric and robot data.