Uber Advanced Technologies Group ("ATG") is returning to Toronto roadways with its self-driving cars in manual mode following the recent release of its self-driving Safety Report and summary of its top-to-bottom internal safety review.
Starting today at 9:00 am ET, self-driving vehicles will begin manual driving with a Mission Specialist sitting behind the wheel controlling the vehicle at all times on city roads and highways near the Company's self-driving research and development hub located in the MaRS building. A second Mission Specialist will sit in the passenger seat. These vehicles will be driven in Toronto to collect data and fuel novel AI technology, automated mapping, and mapless driving research efforts.
Manual-mode driving will allow Uber to collect data needed to understand the different driving conditions that a self-driving vehicle might encounter when operating, and helps ensure the Company's self-driving systems can perform well under various conditions. Building robust systems is critical for success and safety, and capturing the right data is especially important for AI algorithms.
Self-driving systems today require a detailed high definition map of the road in order to safely operate. Creating these maps is a labour intensive effort, requiring repeated manual driving through areas where the self-driving vehicle is expected to travel. Uber expects the manual nature of high definition map-building will slow down the scalability of self-driving technology over time. In an effort to address this challenge, ATG's R&D team in Toronto developed AI technology that enables Uber's self-driving cars to drive where maps are unavailable, building the map in real-time.
"Our automated mapping capabilities, once mature, have the potential to expedite the rate at which we build high definition maps, which is a time-intensive process," said Raquel Urtasun, Uber ATG's Chief Scientist and Toronto office lead. "We are pleased to resume manual driving of our autonomous vehicles in Toronto and excited about the potential that this data collection can have for our self-driving AI research efforts in Canada."
In addition to returning to manual mode on local roadways, Uber is sharing more detail around the Company's plans to grow R&D efforts in Toronto. With the team expected to double within the next year, the R&D hub will relocate to Bathurst College Centre in 2019, a four-storey, commercial mixed-use development owned and operated by RioCan Real Estate Investment Trust, located on Bathurst Street between Dundas and College in Toronto. This move signals Uber's commitment to not only investing in R&D for self-driving, but also the local technology ecosystem for both self-driving development and the core business. Earlier this year, Uber announced the Company will be opening an engineering hub in early 2019 and expanding its ATG Research & Development Centre in Toronto, investing more than C$200 million over five years.
This week, Uber received authorization from the Pennsylvania Department of Transportation to operate as an approved tester of self-driving vehicles in the state. With this approval and following a series of over seventy track verification tests, Uber will resume self-driving operations in Pittsburgh as of today with two Mission Specialists in the vehicle at all times. Furthermore, in a similar manner to Toronto, the Company will also resume today manual driving of its self-driving vehicles in San Francisco.