Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Last year, Huging Face, AI Dev, Lerobot, launched a set of AI open models, data and tools to help build robot systems in the real world. On Tuesday, Huging Face collaborated with Ai Yaak to expand Lerobot with a group of robots and cars that can move in environments, such as the city’s streets, independently.
The new group, Driving learning (L2D) is calledIt exceeds the size, and contains data from sensors installed on cars in German driving schools. L2D captures camera data, GPS (GPS) and “vehicle dynamics” from driving coaches and students who move in the streets with construction areas, intersections, highways and more.
There are a number of open training groups for self -control from companies including Alphabet’s Waymo and Campa Ai. But many of these things focus on planning tasks such as detecting and tracking objects, which require high-quality comments, according to L2D-creators, which makes it difficult to expand their scope.
On the other hand, the L2D is designed to support the development of “comprehensive” learning, and its elements are called, which helps to predict action (for example when the pedestrians can cross the street) directly from the sensor’s inputs (such as camera shots)
Harsimrat Sandhawalia and Remi Cadene, a member of the YAAK, wrote a member of the Robotics team in Huging Face, written in a blog post: “The IQ community can now build self -driving models.” “The L2D aims to be the largest open self -data collection for the source that enables the artificial intelligence community through unique and varied” episodes “to train spatial intelligence from one side.
Face FACE and YAAK are planning to perform “closed loop” tests in the real world of trained models using L2D and Lerobot this summer, and it was published on a car with a safety driver. Companies invite the artificial intelligence community to present the models and tasks they want to evaluate models, such as moving in circles and parking places.