Waymo has now achieved some milestones this year, in the month of August surpassing 10 million real-world miles with its driverless cars and last week launching Waymo One, a commercial driverless taxi service. But its researchers have their eye now fixed on the future.
“In recent years, the supervised training of deep neural networks using large amounts of labeled data has rapidly improved the state-of-the-art in many fields, particularly in the area of object perception and prediction, and these technologies are used extensively at Waymo,” the researchers wrote. “Following the success of neural networks for perception, we naturally asked ourselves the question: … can we train a skilled driver using a purely supervised deep learning approach?”
In an attempt to create a system which is capable of imitating an expert driver, they crafted a neural network, dubbed chaufferNet appropriately, that learned to generate a driving trajectory by simply observing a combination of simulated and real data which includes the map, surrounding object, past monitors of cars and traffic light states. A low-level controller converted the then points trajectory to steering and acceleration commands, which even allows the Artificial Intelligence model to both the digital and real cars.
The model was just a few examples from the “equivalent of about 60 days of expert driving data”, with the techniques that ensured that it did not extrapolate from past motion and actually reacted to changes in the environment. In tests, it even responses to traffic controls such as stop signs and traffic lights, but predictability performed poorly when exposed to situations it would have never seen before.
“Fully autonomous driving systems need to be able to handle the long tail of situations that occur in the real world,” the researchers wrote. “The planner that runs on Waymo vehicles today uses a combination of machine learning and explicit reasoning to continuously evaluate a large number of possibilities and make the best driving decisions in a variety of different scenarios … Therefore, the bar for a completely machine-learned system to replace the Waymo planner is incredibly high, although components from such a system can be used within the Waymo planner, or can be used to create more realistic ‘smart agents’ during simulated testing of the planner.”