We’ll be starting Season 6 of the Smarter Cars podcast in January, with a primary focus on the state of micromobility in Europe. In case you missed any of our Season 5 episodes, here’s a recap with links to the episodes:
Jody Kelman
Director of Product Management, Lyft’s Self-Driving Platform
We discuss Lyft’s two-pronged approach to autonomous vehicles. Lyft has partnerships with Aptiv in Las Vegas and Waymo in Arizona, where partner vehicles are made available with safety drivers on Lyft’s platform as an option riders can choose. And Lyft is also building its own autonomous driving system through its division called Level 5.
Jody discusses why Lyft is pursuing both strategies simultaneously, noting it is part of Lyft’s ethos of working collaboratively to move forward in all directions.
“I think of this as as probably the most difficult problem that my generation has had to solve, and we simply don’t believe anyone is going to be able to solve this by going it alone. So it is going to be this kind of very heavy, three-way collaboration between regulators in cities, industry players and then consumers, who are also very much a part of this adoption story.”
Jody notes that the partnerships in Las Vegas and Arizona are designed to learn how self-driving cars operate on a rideshare platform and how passengers interact with the vehicles. She notes that after a few minutes the ride is just a regular ride and the novelty of self-driving wears off.
“I think those of us who’ve written in self driving cars would say, yeah, they are the best ride your grandmother ever gave you.”
She explains they are not pushing the edges of autonomy with passengers in the car, they leave that for test vehicles, and instead are looking to expose the technology to consumers and get feedback.
As for Lyft’s Level 5 division, Jody notes:
“We weren’t confused that we were a little bit late to this game. But what we knew was that, what we are able to do is really take advantage of what we have that frankly other players in the market don’t, which is a fleet out on the road today.”
Level 5 is testing cars in California and also providing rides to Lyft employees in Palo Alto.
We also discuss issues that ride services can cause in cities, shared rides, and policies to promote multi-modal trips and better traffic flow in congested areas.
Jason Stinson
CTO and Co-Founder of Renovo
We discuss Renovo’s origin story going back to 2010 — from its original electric supercar The Coupe, to working with Stanford on its self-driving DeLorean named Marty, to developing an automotive software platform to help companies manage data from autonomous driving and assisted driving systems.
Jason describes Renovo’s software platform and for OEM companies running ADAS systems it can make their vehicles more similar to Tesla by assisting in gathering, sorting, prioritizing and offloading the relevant data and sending it back to the manufacturer. As Jason notes:
“a really easy way to think about it is our management platform allows you to do things like how Tesla is doing them. It allows you to create a rich management layer on your data itself. It allows you to run applications, workloads on that data in the vehicle, treat the vehicle as an edge processing device, it allows you to transfer the data intelligently, add filters, prioritization, retention policies, all of these things that are required in order to manage these massive amounts of data, which is exactly as you said, this is how Tesla treats their vehicles.”
Renovo provides an end to end data management platform for fully autonomous vehicles and ADAS fleets. Jason notes that the Renovo system allows companies to “manage the data in the vehicle, manage applications that find those nuggets of gold out of that data, and then allows you to take that information off the vehicle. And it’s really important for autonomous vehicles because of how much data they’re generating. And it’s really important for ADAS test development fleets.”
Jason’s a car guy and a tech guy so we also discuss his thoughts on the future of autonomy, whether there will ever be a Level 3 car, and what it will take to get autonomous vehicles deployed at scale.
James Wu
Founder and CEO of DeepMap
We discuss DeepMap’s business and customer model, and the importance of mapping to autonomous vehicles and ADAS systems. DeepMap provides a map engine as a service — using either the customer’s data or its own mapping rig, it creates high definition maps for autonomous vehicles and assisted driving systems in the areas needed. Then it creates a quality controlled map engine service, with maintenance and updates to the maps based on data from the customer’s fleet on an ongoing basis. James notes that the customer’s autonomous or ADAS fleet will have plenty of data but what DeepMap provides is the engine to incorporate and update that data in a high quality way.
James describes the process of building a custom HD map, and how it is different for each customer. He notes there is a base layer like the road, and a geometric 3D layer, and then a “semantic” layer showing the rules of the road like speed limits and traffic lights. He notes that DeepMap can also add customized layers depending on customer needs.
We discuss how high definition maps are created, how mapping and localization work with the autonomous stack, the technical and compute challenges for mapping, and the future of autonomous vehicles post-pandemic.
Warren Logan
Policy Director of Mobility and Interagency Relations for the Mayor’s Office of Oakland, California
Warren works closely with the City’s Department of Transportation, Public Works Department and other Bay Area public agencies to develop strategies that advance the city’s vision for safe and sustainable transportation for everyone. Warren joins us to discuss racism in transportation. We discuss how racism affects street safety and freedom of movement for black people, and the impact of policing and enforcement in public spaces. Warren points out that street safety is much more than avoiding collisions. He notes:
“When you go to community members, especially black and brown communities and say, hi, I’m a bicycle planner and I want to help you bicycle more safely. You’re going to get a conflict of interest there because people are gonna say, well, that’s not the reason I feel unsafe on the streets. I feel unsafe because every time I go outside my house, I’m harassed by police or I’m harassed by institutions that make me feel unwelcome. And that has nothing to do with the color of the stripes on the road and so much more to do with the color of your skin.”
We discuss ways in which transportation departments can re-evaluate enforcement in order to reduce interactions with police that are often dangerous to black people. When I pointed out how messed up it is that we have to exclude police from transportation enforcement due to racism, Warren agreed:
“Sometimes I look at the news actually often, most times when I see the news of yet another black or brown person being killed by the police, I think to myself, like how freaking hard is it to just not do that, right? Like you have options…we have seen interactions between the police and extremely violent people. Like for example, white people that have shot up an entire church or something. And yet somehow they’re able to subdue this person and bring him to a jail instead of killing them. And that’s the part that I think black people especially are just so done with, because it’s like, how hard is it to just enforce the law and then not kill us?”
We also discuss how community engagement and building diverse teams can help serve communities historically left out of urban planning, and how cities and companies need to get out of their silos to address the intersectional issues of transportation, housing, jobs, policing and criminal justice. Warren points out that diverse teams ask the right questions around what problems you are trying to solve:
“You should hire the people you’re trying to help. And on one hand, that sounds so simple, but for some reason it’s complicated for a lot of people. But I think that it’s important to have the people who are building the conversation, be the people who understand where the conversation needs to go.”
Finally, we cover slow streets, bike lanes and quick build projects and the reactions to those infrastructure improvements in many black communities.
Matthew Johnson-Roberson
CEO and Co-Founder of Refraction AI
We discuss the latest developments in robotic delivery with Refraction AI, a company that builds and deploys robotic platforms to provide safe and scalable last mile goods delivery in urban areas.
We discuss Refraction’s current business model, providing delivery services to restaurants and grocery stores in Ann Arbor, Michigan. Matt explains that rather than having a DoorDash type model, Refraction is just the fulfillment layer and the restaurants have their own websites and just choose when to use Refraction for delivery, for a flat fee:
“As opposed to trying to offer this full stack solution, where we do the marketing, we do the advertising, we bring the customer in, we do all of that. We very simply are the fulfillment layer for these restaurants. And so that’s what we’ve been rolling out and over the last month, and that’s what we’re going to try to expand with and our go to market.”
We discuss Refraction’s custom delivery pod that drives autonomously in the bike lane with tele-operation back up. It’s a three wheeled electric bike by design that fits into a bike lane or travels on the side of the road.
We also discuss Refraction’s future plans to expanding into delivery service for other sectors and geographies to become the Stripe/Square of last mile delivery.
“So if you think about things like Stripe and Square, they’ve been really enabling for your local flower shop to be able to offer some type of eCommerce platform and to be able to take mobile payments. And then what if we could do the same thing for actually fulfilling those orders…Square or Stripe doesn’t really care if you’re selling flowers, you’re selling coffee or you’re selling whatever — they’re able to make that work. And we’re hoping that we’re able to do the same thing for fulfillment. Now that’s a tall order. We’re not there yet. I will be the first to admit, but it’s on a roadmap…”
Refraction is looking to expand to other cities which may include Palo Alto or Boston.
Dmitry Shevelenko — CEO and Founder of Tortoise
In this episode, we talk with Dmitry Shevelenko, CEO and founder of Tortoise, a company providing remote repositioning services for lightweight electric vehicles such as scooters and delivery robots.
We discuss the challenges facing the shared scooter industry, and how Tortoise can help solve operational cost and revenue issues by using tele-operation to remotely reposition scooters to prevent clutter, make scooters more available to find and rent, recharge scooters and provide on-demand rental capability.
“So a classic use case for shared micromobility is first and last mile to public transit. And right now, if you’re doing that in a suburban area, and somebody takes a scooter to a train station in the morning, there’s not going to be anybody coming into that train station, till the afternoon/evening rush hour. With real time, remote repositioning, you can return the scooter to a residential area after that first morning trip for a second morning trip, and third morning trip. So through that six to 10:00 AM morning rush hour, you can get four rentals as opposed to just getting one, which is what’s happening right now, and then rinse and repeat in the afternoon and evening as well.”
We discuss Tortoise’s current operations in Peachtree Georgia, and its future plans for deployment. Where possible, Tortoise can provide an on-demand experience where users call for a scooter at their location and the scooter drives to where they are. This is available in Peachtree, but may not work until there are sufficient supply of scooters available in an area.
“We certainly think that on demand is the ideal experience, but depending on how dense the environment is and how many total scooters are allowed, and how many Tortoise compatible scooters are there, you might want to start with using us for demand-side repositioning that anticipates where you’ll have demand, as opposed to, going for the full Uber experience on day one.”
We also talk about how repositioning can help with scooter recharging, since the scooter can be repositioned to a charging station when low battery instead of waiting for human teams to pick up the vehicles for charging at the end of the day.
“Right now operators are very constrained in terms of how frequently they can recharge. With the current fleet team model, or with gig workers, they’re only doing the recharging during set times a day. So say you have somebody take a scooter up a really big hill, and you get two of those types of trips in the morning, and the scooter is already depleted 90% of its battery. You’re stuck with that scooter just sitting out in the street, all day long, not really being a rentable, until it gets picked up. Part of where the higher utilization and the recharging overlap is being able to do a midday swap or recharge of a scooter that has a low battery.”
We also discuss Tortoise’s entry into the delivery market with a sidewalk robot making deliveries in a cart that can carry groceries, liquor or other goods up to 150 pounds at a max speed of 7 mph. Subsequent to our podcast, Tortoise revealed a video of its delivery vehicle which you can see here.
Modar Alaoui — CEO and Founder of Eyeris
In this episode, we talk with Modar Alaoui, the CEO and founder of Eyeris, a company that provides in-vehicle scene understanding for cars with ADAS systems and fully autonomous vehicles.
In-vehicle scene understanding can be used for tasks such as driver monitoring, occupant monitoring, safety and convenience adjustments and personalization, and determining whether a fleet vehicle needs to be cleaned or if items were left behind by a passenger.
Modar notes that Eyeris is different from typical driver monitoring systems which only focus on eye tracking:
“Historically when the automotive industry started thinking about including driver monitoring, they wanted to focus more on one feature that is eyes on- road, eyes off-road. And therefore, clearly the focus of the algorithms are eye tracking…. And so historically eye tracking has been around for a long time and to the point where it has achieved a very high level of accuracy, however, the methods used at that time, traditionally known as a corneal reflection, are very rigid. So they require a camera to be placed, in a frontal position to the face, they have a narrow field of view, they require LED or some sort of active illumination that can synchronize with the camera shutter, or they require a special type of camera. And so, all of that was actually done initially by a couple of automotive OEMs until they realized, a year or two later, that although it offers the highest level of accuracy, the degrees of freedom were very limited.”
Eyeris has developed algorithms that evaluate and understand human behavior, surface classification and object localization using cameras and other sensors and an extensive data set built over many years. This provides broader scene understanding than just a driver monitoring capability.
Graham Gullans — VP of Business and Corp Dev, Superpedestrian
In this episode, we talk with Graham Gullans, VP of Business and Corporate Development at Superpedestrian, a company that makes intelligent electric scooters and bikes and now operates a shared micromobility service called Link.
Superpedestrian has created a scooter with advanced vehicle intelligence that can perform self-diagnostics to find issues before they affect a ride. This allows the fleet to have more up time and they can avoid having scooters on the street that are not ready to ride. Graham notes:
“Every 30 minutes, our scooters go through a safety check when they’re not in a ride, and then every 30 seconds when in a ride, they go through a safety check and all that data gets sent back to us to make sure we know what’s happening with the vehicle. The safety check is primarily to make sure the mechanical and electrical systems are to the point where they’re safe for a rider to get on, because we want you to know that when you get on that scooter, it’s safe to ride.”
We discuss the use of vehicle intelligence to improve safety, reliability and compliance for shared scooters and how Superpedestrian is taking on some of the biggest challenges that shared micromobility operators face. Superpedestrian has created a sturdy, larger scooter with a bigger battery to create a smoother ride that can go further on a single charge and improve operations:
“Early in 2018, we realized that it’s labor intensive to pick up scooters every day or twice a day to recharge them…So we stuck a gigantic battery in our vehicle…an 84 cell battery in the platform, that’s nearly one kilowatt hour, which is by far the largest of any scooter that’s out there. If you rode it from point A to point B without stopping you would get about 55 miles or, five hours, on the scooter.”
Superpedestrian is also focused on compliance with city regulations with geofencing capabilities that can react more quickly to identify scooters that are traveling in areas where scooters are not allowed.
Evangelos Simoudis — Managing Partner, Synapse Partners
In this episode, we talk with venture capital investor Evangelos Simoudis, Managing Partner of Synapse Partners and author of two books on the future of transportation — The Big Data Opportunity in Our Driverless Future and his newly published Transportation Transformation.
We discuss his views on the future of autonomous vehicles, ride-hail services, micromobility and public transit. We explore his thesis that next generation mobility — combining intelligent infrastructures provided by cities, with multi-modal transportation and goods delivery services — will require new value chains with novel business models and will rely on the collaboration of three constituencies — the automotive ecosystem, mobility services companies and governments.
Jewel Li — COO of AutoX
We talk with Jewel Li, the COO of AutoX, a company making an autonomous driving system for cars, delivery vehicles and long haul trucks.
AutoX recently was the third company to receive a permit to test autonomous vehicles in California without a driver in the car. Jewel discusses the difference between the US and Chinese AV markets:
“I think when you look at the landscape of the industry, there are a lot of similarities between the US and China markets, but the difference is also huge. The similarity is that, for Level Four, the most valuable markets in both countries are the same, the biggest use cases that is the robo-taxi, the robo- delivery truck and the robo-truck. But the difference in the landscape is that the competition in the US is a lot more extreme….There are a lot of competitors in the U S market, we can easily name 10 of them, which have raised a lot of funding and have big operations. But when you look at the China market, the competition is a lot less fierce. There are some bigger companies like Baidu and Didi. These are the only two major companies that have a tech background. The others are startups like ourselves and also Pony which got investment from Toyota, and very few others. So the competition fierceness is different.”
We discuss AutoX’s technology and business model, the driverless permit process in California, the testing that AutoX is doing in China, the differences in testing, regulation and infrastructure between China and the US, and AutoX’s partnerships and ride service pilots.
Sam Kansara — Senior Product Manager, Waymo
In the Season 5 finale, we talk with Sam Kansara, a senior product manager at Waymo who is scaling the Waymo One driverless ride service and Waymo Via for trucking and local delivery.
Waymo has been the leader in developing autonomous driving technology and the first ride service to remove the safety driver for a portion of the rides it offers in the Phoenix area. As far as expanding that operational design domain into other areas, Sam notes the cautious and safe approach taken by Waymo to new areas:
“We’ve learned a lot in these last 10 years. And one thing that we’ve definitely learned is that expanding our product is going to be gradual. It’s going to take time and it’s going to be one that we want to make sure we do in a slow and a measured way.”
We discuss Waymo’s ride service, how Waymo thinks about safety and deployment, Waymo’s Via delivery service and the user experience and expectations around offering fully driverless rides.