Closing the Lab-to-Store Gap: A Data-Efficient Post-Training and Experience-Driven Learning VLA Framework for Retail Humanoids
Researchers introduced DEED, a new framework designed to help Vision-Language-Action models handle real-world retail tasks more efficiently.
- DEED framework addresses the 'lab-to-store' gap for humanoid robots.
- Tested successfully on Unitree G1-Edu robots performing retail restocking.
- Uses GR00T N1.6 foundation model as a base for VLA tasks.
- Focuses on data efficiency to reduce the cost of real-world training.
The transition from controlled laboratory settings to unpredictable real-world environments remains a significant hurdle for Vision-Language-Action (VLA) models. Current humanoid robots often struggle with execution errors and environmental shifts when moved from simulation to physical stores.
To address this, researchers developed the DEED framework, which focuses on data-efficient post-training and experience-driven learning. The system was tested using a Unitree G1-Edu humanoid robot performing chip-restocking tasks in a supermarket setting.
By utilizing the GR00T N1.6 foundation model, the framework employs a specialized pipeline that aligns control frequencies and improves how robots learn from physical experiences. This approach aims to make humanoid deployment in commercial spaces more reliable and less reliant on massive, expensive datasets.
Provides a new methodology for post-training VLA models for physical hardware.
Offers a potential path toward deploying humanoid robots in retail and logistics.
Represents a significant research advancement in robotics and foundation models.
- Vision-Language-Action (VLA)
- A type of AI model that processes visual and textual inputs to output direct physical actions.
RoboticsHyundai claims humanoid robot plan is not part of talks with striking workers
Courteous Anticipation: Improving Long-Lived Task Planning in Persistent Shared Environments
RoboticsUkrainian drones deliver robots directly into battle by sea and air
From Distances to Trajectories: Real-Time Signed Distance Function Mapping and Distance-Accelerated Motion Planning for UAVs
RoboticsGritt exits stealth with $34 million for robots to build solar plants—then, everything else
Artificial Intelligence in Aerospace: Closing the Delivery Gap - Tata Consultancy Services
Tata Consultancy Services discusses the role of artificial intelligence in closing the delivery gap in the aerospace industry. This involves leveraging AI to improve efficiency and reduce delays.
HardwareNVIDIA AI Supercomputer Comes Online at Naval Postgraduate School
NVIDIA has officially commissioned a DGX GB300 AI supercomputer at the Naval Postgraduate School to support military research and education.
A quantum mechanics approach to artificial intelligence can improve cancer outcomes - Technology Org
Researchers are exploring the use of quantum mechanics in artificial intelligence to improve cancer treatment outcomes. This approach has shown promise in enhancing the accuracy of cancer diagnosis and treatment.
Wake Forest launches AI for Human Flourishing initiative; appoints William Fleeson as Associate Provost for AI Initiatives - Inside WFU
Wake Forest University has launched an AI for Human Flourishing initiative, appointing William Fleeson as Associate Provost for AI Initiatives.
BusinessAfter shocking quarter, IBM insists that AI isn’t killing the mainframe
IBM leadership claims that recent shifts in corporate spending toward AI are only temporarily impacting mainframe sales.
Michigan Tech Atmospheric Scientists Linked to Three Research Projects Selected for DOE Genesis Mission - Michigan Technological University
Michigan Technological University's atmospheric scientists are part of three selected research projects for the DOE Genesis Mission.