Introducing the Path to AV #PathToAV
I chase self-driving cars on the roads of California where sixty self-driving car pilots share the road with us. The fascination started for me many years back as I noticed that the Google self-driving car (now Waymo) shared El Camino Real in Mountain View every night as I drove home after teaching Cognitive IoT course at Stanford.
I pondered about our class discussions about connected devices and AI from sensor data and about what is the business problem and who is the customer for these expensive self-driving car pilots. It made me question what was the self-driving car actually testing. Was the car testing to run as a robo-taxi? No, it was not treating anyone as a customer and just wanted to navigate the roads safely. Was the car testing how to share the road with human drivers like me to get ready to launch an autonomous vehicle anytime soon? Obviously not, it was immune to me sharing the road every day on the same path. There was no signs of what the car thought of us as consumers or a path to full autonomy to share the road in some definite timeline. Yes, I did think about what the car was thinking. Thats another story for another day.
ok, so the self-driving car was like a toddler learning to walk, self-centered to make sure it did not crash onto anyone. It was piloting to test its technology, car cognition, to safely understand the drivable part of the road, stay within the lane, identify traffic signs, and recognize everyone and everything else from other cars to pedestrians as objects to be avoided. So it became clear. I was just training data for the computer vision models of the self-driving car. I made a decision then and there!
If I was training data, I was going to train the self-driving car well and be the best training data on earth.
So began my obsession to meet, chase and train every self-driving possible. I watched how self-driving cars reacted in different road situations, navigated constuctions zone and how they drove defensively and avoided risky situations. I have videos of every make and model and several homegrown self-driving cars and two versions of Apple self-driving cars and a Waymo truck from California. I have visited many places globally just to meet the AVs there. (Ok thats not true. I did visit Barcelona to checkout their smart city connected garbage bins) I am still waiting to meet some of the self-driving cars from other parts of the world. In summer of 2018, I went across Boston multiple times for a chance to participate in a self-driving Lyft pilot with Aptiv. I visited Surrey, UK to spot ZEBRA self-driving car from Prof. Aldo Sorniotti and met with the Catalonia team that tested the self-driving vehicle in Spain. That is where my friend David Kerrigan from Ireland began teaming up with me to compare the self-driving cars he has met. He visits US to speak at my Standord Continuing Studies Autonomous Bootcamp class each year and we watch self-driving cars together.
Technology of Self-Driving Cars
I have dug through every possible material and learned how self-driving car works. There are 4 steps that happen in self-driving cars in split seconds. The car first localizes itself, then it ‘sees’ the road from its sensor data and identifies the lanes and objects all around it. Then, it does path planning to decide to move forward or turn left or right or stop or give control to the human driver. The final piece is by which it gives control to the drive-by-wire mechanical parts such as brake or accelerator to execute on the planned path. Autonomous driving is a trail and error of sensor fusion, the intelligence of the AI inference model version in the car and the AV operator watching for disengagement to take control.
You can look at Udacity student’s github (its about 3 years old but good to play with) to learn lane detection from deep learning steering model . Definitely listen to Prof. Lex Fridman’s self-driving course archive from MIT to learn deep learning models and meet fellow technologists learning to build the tech to get us to Level 5. (where the car can drive fully autonomously better than us humans who cause 6 million car accidents on US roads alone per year)
Vatsal Srivastava does a wonderful job of explaining how a self-driving car is built . (See reference below)
I spent the holiday season of 2015 digging into disengagement reports of then 30+ companies who submit a report to DMV about why a self-driving car gives control back to humans every time it drives. Whether Disengagement is a valid metric of how advanced or safe is a self-driving car’s technology is left to debate of AV regulators. I ended up writing a book called ‘2030 The DriverlessWorld” as a time-traveler narrative on the business models, innovations and regulation needed to get us to making driverless cars mainstream. I came up with 2030 as a realistic time frame to expect autonomous vehicles to be generally available (when I wrote the book in 2016).
I wanted to find out why 60 companies are racing to get us autonomous vehicles now and as always my secret agenda is to find out how my students can innovate in this space.
Enter Tesla and Innovation Labs of OEMs
That day in April 2017 when Tesla’s market cap (valuation) crossed GM’s for the first time is what I would name as the birth of Autonomous Vehicle industry. By then Waymo Team had spent 8 years piloting self-driving cars in consumer roads while John Deer tested autonomous tractors in precision agriculture in farms in Germany to take the lead on AV Technlogy. But the electric car built by Tesla Motors challenged the status quo and began treating the car as software on wheels disrupting the entire industry. Tesla’s promise of Autopilot is a reality today and the debate rages on that whether Tesla’s autopilot is true Autonomous Vehicle technology which they offer as a consumer chosen add-on to skirt the regulation requirement for self-driving car pilots. And the race began.
Car manufacturers (OEMs) who have been on a decade long journey prior to that to digitize the car were focused on infotainment systems with proprietary integration with Tier1 systems. They built innovation centers in Silicon Valley and tested out business models for dealers, trucks, buses, and mobility-as-a-service models for transportation. The OEMs invested, partnered and acquired self-driving car technology startups. Maven by GM, Argo.ai by Ford, Lyft’s partnership with Aptiv and Mercedez-Bens’s investment in Via to replace city buses with on-demand mobility all expanded the market to endless opportunities. And it fed the imagination of innovators in these companies and my students!
A little about my students
tl:dr: My students are amazing. My students are professional adults who love technology, innovate in their jobs and are willing to find time to learn and go past the intimidation of entering a new field. They work as innovation managers, product managers, business managers, or engineers who are entrepreneurs wanting to understand how to build out the business side of technology. They come to my class to hungry to change the world. They make me smart. They enjoy the discussions where we go into the gap between the depths of technology and high level business strategy.
And my class speakers!
They come from all over the world Oliver Cameron showed the first functioning robot-taxi giving mobility to a visually impaired woman in a senior community in San Jose. Roxy Stimpson showed us the power of sensor data to build a business solution. Jane Ren, ceo of @atomiton showed us the power of digitaltwins and got us thinking about digital twins in case. My class guest speakers come to share the reality of the industry and have fun in an engaging discussion with hungry, smart people looking to connect the dots from their careers to the AV space.
Data in the Car
It might be my background in Growth analytics and love of data or just the nature of connected devices, that my research led to the amazing world of data in the car. Car Cognition is powered by self-driving cars. We are the data that trains the car cognition. Self-driving car companies have released open AV datasets for you to improve car cognition. I recently did a session about AV Data in my weeklywed YouTube #CareerPivot session and showed car data from public datasets.
It is fun to look at this data for product managers to understand different ways of labeling training data, for designers to know what different data is available to create in-car and intra-car experiences and for business managers to know what data really means in an ecosystem to build out a car into a mobility service or an autonomous vehicle solution for a particular industry.
Mobility of data is what creates value in an ecosystem. Something to think about if you share my excitement for car data.
We have barely scratched the surface on the Path to AV!!!
We haven’t spoken about regulation catching up to innovation in AVs, the business drivers and ownership shifts or business models that create data driven disintermediation in AV ecosystem expanding past automobile to retail, healthcare, mining, transportation, cargo delivery, logistics and ….. lots more.
Are you wondering why the autonomous vehicle space is a lucrative career space now as we get ready for a post-covid world. Share your thoughts or
Introducing Path to AV
It is time we go from individual courses to learn about AV business, and data to pivot your career to Autonomous Vehicles with a Path to AV #PathToAV series of courses on DriverlessWorldSchool.
Come learn, network, understand the ecosystem and develop real industry experience with a capstone project on you career pivot on a #PathtoAV.
Introducing the DriverlessWorldSchool Mentors
- David Kerrigan is the top writer about self-driving cars here on medium. He is going to teach the AV Masterclass, a 6 weekly 1 hour live online course.
- Alexandre vargha of Volvo group brings 15 yrs experience from the OEM world to teach DigitalTwins in the car in the AV Data Course.
- Dr. Charles Ikem, the brilliant AI Designer and Service Design expert will teach AIX, Product Design of in-car and intra-car experience. He is planning to offer AIX for MachineLearning, AR, ComputerVision and Voice as a business hackathon experience instead of a typical online class.
Talk to us, we want to be with you all along the way on your journey on your #PathToAV.
Join me and David Kerrigan and my AV students on a LinkedIn chat #PivotToAV on wednesday 10am pt/19 cest, right after ur #weeklywed #careerpivot #live. You can join by adding a question or tagging a friend to the post below.
Tell me what excites you about Autonomous Vehicles? Where are the opportunities? What are the challenges to get to full autonomy? What is the role of regulation? What industries are impacted? What scares you most about AV as a career? If you are my past student, please join to share your experience in AV space.
If this topic is useful, clap, share and tell your friends. Tell me what topic in AVs should I write about next and I am happy to make it a series.
References:
- Car accidents stats from wikipedia
- MIT’s Prof.Fridman’s Self-driving Car lessons and community
- Udacity’s self-driving car github from Eric Gonzalez Oliver Cameron
@macjshiggins (be sure to follow them). - Technology of self-driving car explained
- AV Data from Open datasets from DriverlessWorldSchool
- PivotToAI foundation course series
About Sudha Jamthe:
Sudha Jamthe is a technology futurist, Stanford CSP teacher who researches, keynotes, writes books and teaches with a passion to mentor business leaders to innovate in technology ecosystems to create the driverlessworld. She has an unquenched thirst to find answers on how to monetize data and new business models. She teaches at Stanford Continuing studies and online at DriverlessWorldSchool. You can catch her at her weeklywed YouTube #CareerPivot session or on Twitter.
