310 papers - avg viability 6.4
Current research in robotics is increasingly focused on enhancing the adaptability and efficiency of robotic systems in dynamic environments. Recent work on force-adaptive reinforcement learning frameworks is enabling humanoid robots to maintain balance and manipulate objects under varying external forces, which is crucial for real-world applications. Simultaneously, advancements in motion generation are addressing the challenges of high-dimensional robots, allowing for safe and efficient trajectory planning that significantly outperforms traditional methods. The integration of visual-language-action models is also gaining traction, facilitating multi-task learning and skill acquisition without the need for extensive retraining. Moreover, innovations in teleoperation, such as lightweight haptic feedback gloves, are improving the quality of human-robot interactions, which is vital for tasks requiring dexterity. These developments collectively aim to solve commercial problems in areas like manufacturing, logistics, and autonomous navigation, making robots more capable and versatile in complex, real-world scenarios.
FAME is a force-adaptive reinforcement learning framework that enhances humanoid manipulation by adapting to external forces in real-time.
A low-cost, cable-driven force feedback glove for dexterous teleoperation that significantly improves task success rates and reduces completion time, with open-source code and designs available.
A panoramic Visual-Inertial SLAM framework with a novel real-world UAV dataset, offering superior accuracy and robustness for drone localization and mapping.
An end-to-end neural network enables quadrotors to navigate cluttered environments at high speed using event camera data, trained efficiently with approximate imitation learning.
Scalable autonomous humanoid skills using trainable guided diffusion models trained on diverse motion data.
C$^2$-Explorer is a decentralized multi-UAV exploration framework that significantly reduces exploration time and path length by using a connectivity graph and contiguity-driven task allocation, ready for commercial drone applications.
ExpertGen automates expert policy learning in simulation for scalable sim-to-real transfer in robotics.
Ruka-v2 is an affordable open-source humanoid robotic hand with advanced dexterity and wrist mobility enabling efficient robot learning.
AtomicVLA enables robots to learn and execute long-horizon tasks by decomposing them into atomic skills, offering a scalable solution for continual learning in real-world environments.
A neural control barrier function method for safe robot navigation in dynamic environments, demonstrated on both ground robots and quadrotors, offering improved success rates.