Waymo, the driverless transportation arm of Google parent Alphabet, has launched a new AI research model for its self-driving operations.
In two press releases on his approach to AI and its new end-to-end multimodal model for autonomous driving, nicknamed EMMAWaymo shared details about its plans for the AI research model in the future. The company says it is still using the EMMA model in the research stage, rather than in operational vehicles, and that this approach is an alternative that closely resembles Tesla’s Fully Self-Driving (FSD) and other approaches. end-to-end model.
“EMMA is research that demonstrates the power and relevance of multimodal models for autonomous driving,” said Drago Anguelov, vice president and head of research at Waymo. “We are excited to continue exploring how multimodal methods and components can help create an even more generalizable and adaptable driving stack.” »
Waymo says the EMMA model uses real-world knowledge based on its Gemini language model, while the end-to-end approach should eventually enable autonomous vehicles to operate directly from data from sensors and driving scenarios in real time. The company also highlighted its use of extended language models (LLM) and vision language models (VLM), calling its architecture the Waymo Foundation model.
Hear company executives detail Waymo’s research and AI program below.
EMMA Research and Reviews
In the EMMA press release, Waymo outlines the following as key aspects of the research program:
- End-to-end learning: EMMA processes raw camera inputs and text data to generate various driving results, including planner trajectories, perception objects and road graph elements.
- Unified linguistic space: EMMA maximizes Gemini’s knowledge of the world by representing non-sensor inputs and outputs as natural language text.
- Chain of thought reasoning: EMMA uses chain-of-thought reasoning to improve its decision-making process, improving end-to-end planning performance by 6.7% and providing interpretable justification for its motor decisions.
“The problem we’re trying to solve is how to create autonomous agents that navigate the real world,” says Srikanth Thirumalai, Waymo’s vice president of engineering. “This goes way beyond what many AI companies are trying to do.”
Still, some have questioned the large-scale end-to-end model, saying it might be too risky to use generative AI models without including meaningful safeguards.
“This is a movement around something that sounds impressive but is not a solution,” said Sterling Anderson, chief product officer of Aurora Innovation, in a statement to Automotive News.
Mobileye CTO Shai Shalev-Shwartz called end-to-end approaches a “huge risk,” especially when it comes to verifying the decision-making process for vehicles running on the model. It should also be noted that Waymo is currently only investigating this approach and has no current plans to make it commercially available.
The news comes later Waymo recently closed a $5.6 billion funding roundeffectively bringing the company valuation exceeds $45 billion. The company is also working on its next generation of autonomous vehicles based on the Hyundai Ioniq 5, built at a new factory in Georgia.
What are your thoughts? Let me know at zach@teslarati.comfind me on @zacharyviscontior send us advice to tips@teslarati.com.
Need accessories for your Tesla? Discover the Teslarat Walk :
