
Google’s self-driving vehicle department Waymo has made strides on its own with the end goal of having a self-driving car on the road as soon as possible. It’s latest acquisition – a British-based AI firm called Latent Logic – might help it achieve this goal sooner than expected. Latent Logic is a spin-off from Oxford University. The company has dedicated itself to creating some of the most realistic simulations of cars and roads that manufacturers can use to train AI drivers about complex human driving behaviors.
Complex Road behavior Training
Waymo aims to utilize the technology that Latent Logic offers to teach its AI drivers to respond to unpredictable events. AI’s ability to deal with unpredictability is notoriously poor and Waymo hopes that by watching and learning, their AI driver will become better at doing so. The company has faith that the simulation will provide enough data for the AI to learn about how it should handle behaviors like another driver cutting into traffic in front of it, or how it could handle a cyclist skidding in front of the vehicle due to rainy weather.
Reinforcement Learning the Key
Latent Logic has perfected the art of imitation learning. The methodology teaches AI behaviors based on its observation of humans performing the same actions and attempting to replicate them. Waymo anticipates that including this learning into its AI training will help it adapt to real-world conditions sooner and be able to handle driving outside of controlled circumstances.
Information About the Deal Not Public
Neither Waymo nor Latent Logic has disclosed details about the deal, such as the amount of money paid to acquire the AI firm. However, this purchase is Waymo’s first foothold in the European and UK markets as far as engineering and tech development. The engineering and technical team will shift location to Waymo’s headquarters to help them with further development, while the company itself remains in Oxford. Both companies have the aim of developing safe self-driving vehicles for consumer use, and the average buyer would benefit significantly from having an AI driver that can predict human behavior.