Momenta Releases First Mass-Produced Intelligent Driving Large Model

TapTechNews September 29th news, the self-driving company Momenta's official WeChat account announced tonight that during the Global Intelligent Automobile Industry Conference, Momenta released the first mass-produced intelligent driving large model, which integrates perception and planning into one large model, forming an end-to-end, deep learning-based self-driving solution.

The official said that Momenta's mass-produced intelligent driving large model can efficiently solve the long-tail problem of self-driving (TapTechNews note: usually refers to special situations or rare events with low occurrence frequency but a wide variety of types), and supports accurately predicting the passing intention of vehicles or pedestrians in various complex road environments, automatically adjusting the vehicle speed, flexibly changing lanes, and calmly avoiding obstacles.

Momenta also claims that this large model can calmly deal with complex intersections or dynamic cross-scene scenarios, which can significantly improve driving safety and traffic efficiency. Even in extremely narrow parking spaces at night, dead-end parking spaces and other extreme scenarios, it can also achieve accurate parking. Currently, this technical solution has been mass-produced and delivered on multiple automotive brands.

According to TapTechNews' previous report, Momenta's chief executive officer, Cao Xudong, once proposed the Moore's Law of Intelligent Driving: the software experience is improved by 10 to 100 times every two years; the hardware BOM cost (raw material cost) is reduced by half every two years.