Technical Directions
Robotics coverage relating to sim-to-real.
204 selected reportsLast 30 days 167 recordsAll coverage 851 records
Hugging Face introduces SmolVLA, a 450M parameter open-source Vision-Language-Action model for robotics, trained on community-shared datasets. It outperforms larger models in simulation and real-world tasks, supports asynchronous inference for faster response, and is designed for deployment on consumer hardware.
SmolVLA is a compact, efficient Vision-Language-Action (VLA) model designed for affordable robotics, trainable on a single GPU and deployable on consumer hardware. It matches the performance of larger VLAs through community-driven data, with a reference implementation and training instructions provided.
Sanctuary AI has demonstrated industry-leading sim-to-real transfer of dexterous manipulation policies using their high-degree-of-freedom hydraulic hands. The company leverages NVIDIA Isaac Lab, an open-source framework for robot learning, to simulate dexterity-focused training environments. This approach enables efficient training of autonomous robots, with the ability to train thousands of hands simultaneously and accelerate the learning process through online reinforcement learning.
NIST is launching the Agile Robotics for Industrial Automation Competition (ARIA) to improve robot agility in manufacturing. The competition, a joint effort with the IEEE Conference on Automation Science and Engineering, will focus on tasks such as failure recovery, automated planning, and plug-and-play robot integration. The goal is to reduce programming costs and increase adaptability for small and medium manufacturers.
Article dates follow your selected timezone. Briefing editions use Hong Kong time (UTC+8).