Building a Better Driver #TCRobotics

Building a Better Driver with Sterling Anderson (Aurora) and Raquel Urtasun (Uber)

Autonomous vehicles can’t be as good as human drivers. They need to be better. Aurora co-founder and CPO Sterling Anderson and Uber ATG Chief scientist Raquel Urtasun dig into the self-driving stack and how AI is used to help vehicles understand and predict what’s happening in the world around them and make the right decisions.

In February, 2019, Aurora raised $530 million in a Series B.

Aurora has multiple AI systems working together on self-driving.

They discussed creating a safety layer in the AI.

Raquel talks about safety at many levels. Safety first with simulation before the test track. There are many levels and verifications before getting it onto the road.

They test and simulate the whole stack.

They are using AI to do a full simulation. AI guides what to simulate.

Just random tests is not enough there is the for more adversarial tests.

Aurora talks about the policy and procedures and sensor modalities.
The policy is to not penetrate a safety zone around other objects.
No data left behind. All data has to be classified from all sensors.

On the road testing is to more rapidly test the off-road simulations.

They permute the data in order to amplify the on road testing.

Backgound

Raquel Urtasun is the Head of Uber ATG Toronto. She is also an Associate Professor in the Department of Computer Science at the University of Toronto, a Canada Research Chair in Machine Learning and Computer Vision and a co-founder of the Vector Institute for AI. Prior to this, she was an Assistant Professor at TTI Chicago. She was also a visiting professor at ETH Zurich during the spring semester of 2010.

She received her Ph.D. degree from Ecole Polytechnique Federal de Lausanne (EPFL) in 2006 and did her postdoc at MIT and UC Berkeley. She is a world leading expert in machine perception for self-driving cars. Her research interests include machine learning, computer vision, robotics and remote sensing. Her lab was selected as an NVIDIA NVAIL lab.

She is a recipient of an NSERC EWR Steacie Award, an NVIDIA Pioneers of AI Award, a Ministry of Education and Innovation Early Researcher Award, three Google Faculty Research Awards, an Amazon Faculty Research Award, a Connaught New Researcher Award and two Best Paper Runner up Prizes awarded at the Conference on Computer Vision and Pattern Recognition (CVPR) in 2013 and 2017 respectively.

Sterling Anderson is the co-founder and Chief Product Officer of Aurora, the company delivering the benefits of self-driving technology safely, quickly, and broadly. A longtime developer of autonomous vehicle technology, Sterling developed the MIT Intelligent Co-Pilot, a shared autonomy framework that paved the way for broad advances in cooperative control of human-machine systems. In 2014, he joined Tesla, where he lead the design, development, and launch of the Tesla Model X, then led the team that delivered Tesla Autopilot.

Sterling holds several patents and over a dozen publications in autonomous vehicle systems, and an MS and PhD from MIT.

SOURCE -Live coverage of Building a Better Driver at TechCrunch Robotics+AI 2019, Aurora on Youtube
Written By Brian Wang, Nextbigfuture.com

2 thoughts on “Building a Better Driver #TCRobotics”

  1. It should be that humans have no choice when driving a vehicle; look at the energy the brains consume while driving a vehicle, is this a choice? Self-driving cars will be a thing of the past, very soon after air and space laws culminate in minimal idistance flight travel. When mankind has colonized the Solar System, (that’s the name) earth’s surface will be nothing more than product transport, large vehicles and rail transport for product/merchandise logistics, passenger vehicles will be outlawed on the surface.

    Coanda Effect – Formula 1 Dictionary

    Reply
  2. I’m afraid self driving will be the victim of a damned if you do, and damned if you don’t dilema. Human drivers have already exhibited dislike of them for their rigid adherence to traffic laws, and their risk adverse driving style. Not only that, self driving cars get rear ended more than humans, likely for the same reasons. Consider what is said about older drivers, because their poorer vision, slower reflexes, fear of losing their license, and fear of injury lead them to develop similar habits.
    These driving characteristics could lead to political difficulties in state, and federal legislatures, particularly in states with little connection to the automotive, or trucking industries. In states where they are allowed provisionally, final permission might be denied, or in states where they gain permission, it might be revoked as human resentment rises.
    On the other hand, allowing AIs to bend the traffic laws, and be less risk averse, in other words have them drive more like humans, could lead to them getting traffic tickets, and being held responsible for accidents they actually do cause. Liability for their owners, and their manufacturers could be huge.

    Reply

Leave a Comment