Waymo integrates a special AI into its autonomous vehicles

Since Waymo is a subsidiary of the giant Google, it was suspected that its autonomous driving technology would benefit from the expertise of its parent company.

Waymo has taken a significant step forward in the field of autonomous driving. The company has always considered its collaboration with Google DeepMind as a competitive advantage, and today we understand how it plans to leverage Google’s multimodal large language model, Gemini, to improve the efficiency of its autonomous taxis.

Waymo introduces a comprehensive multimodal end-to-end model, called EMMA, which can process sensor data to generate trajectories for its autonomous vehicles. Essentially, this means Waymo’s vehicles can make smarter driving decisions and effectively avoid obstacles.

Traditional autonomous driving systems usually develop specific modules for each function, whether perception, mapping, anticipation, or planning. The data are then cross-referenced. Its limitations are obvious, especially when dealing with new complex environments. Waymo believes that since Gemini can solve these problems because it performs a “chain of thought reasoning” that mimics human logical reasoning.

The EMMA model has been developed to help Waymo’s autonomous taxis navigate complex environments. For example, when encountering animals or unmarked construction, EMMA can assist the vehicle in finding the best routes by tapping into its AI data to determine the optimal response.

The match is thus decided, and the Waymo / Google duo takes the lead over Tesla / FSD model? Well, not exactly, because Waymo admits that EMMA has certain limitations, such as its current inability to correlate 3D sensor inputs with radar and other Lidar… In short, the road ahead is still long.

READ ALSO: Waymo: autonomous taxis will operate in San Francisco