As Sheldonl, we're shaping the future of electric vehicles with flexible circuit technology that enables better performance and design freedom in compact spaces.

Visit Sheldonl.com and challenge us to help you shape what's next.

Hi everybody, welcome to Shift, a podcast about mobility.

I'm Pete Bigelow, your host and reporter at the Automotive News.

Educate Scen rare events, needles in the haystack.

Whatever you want to call them, uncovering the anomalous things that happen on the road

amid millions of miles of mundane driving

has been one of the central challenges for companies developing self-driving technology.

Those companies spend substantial amounts of time and money deploying fleets for real world testing

so they can capture data and train their systems.

Or they do that in simulation and run thousands of permutations of every conceivable scenario.

The trouble is, with either or both approaches, the work is never done.

There's always something new and unusual. It turns out it's a pretty regular part of driving.

And practically speaking, when we look at the problems with self-driving cars and

countering emergency responders in San Francisco, for one example, the early results

are not exactly inspiring confidence and trust from the public and public officials.

My guest today is Christine Moon, president and co-founder of Blue Space

AI. She says there's a better way that you don't need to rack up millions of miles of

testing to accumulate training data.

That you don't need to fall into a trap of classifying every object you see on the road.

And that you can have a versatile system capable of operating on the road, off road in truck yards,

i.e. any environment you want. And that you can have a system that's not a black box,

...