#1 cleantech news, reviews, & analysis site in world. Support our work today. The future is now.


Autonomous Vehicles

Published on August 12th, 2018 | by Cynthia Shahan

0

Autonomous Driving Forecasts & Challenges — Interviewing BNEF’s Head of Intelligent Mobility (#CleanTechnica Video)

August 12th, 2018 by  



Zachary Shahan, CleanTechnica Director & Chief Editor, conducted a brief interview with Ali lzadi-Najafabadiat, Head of Intelligent Mobility at BNEF, at The Mobility Conference — a conference earlier this year in Abu Dhabi, UAE, organized by Global EVRT, CleanTechnica, and Masdar. The interview brings us some highlights about and insights into our autonomous future.

Zach leads the discussion with expectations by 2025, inquiring about a point Ali lzadi-Najafabadiat expressed on the panel discussion. In particular, the issue is the necessity of 5G internet for self-driving passenger vehicles.

Autonomous Driving Forecasts & Challenges — BNEF Head of Intelligent Mobility

During Global EVRT and CleanTechnica's first rendition of The Mobility Conference, a part of Abu Dhabi's annual World Future Energy Summit and Abu Dhabi Sustainability Week, CleanTechnica Director Zach Shahan spoke with Ali Izadi-Najafabadi, Bloomberg New Energy Finance's Head of Intelligent Mobility, about autonomous vehicle progress, challenges, and forecasts. Listen in.

Posted by CleanTechnica on Saturday, August 4, 2018


The essentials: Synchronicity, safety, and success in self-driving mobility will arrive through precise processing of data in real-time. Zach and Ali discuss the particulars and differences between some of the main players — Waymo, Uber, and of course Tesla.

Waymo is adding 25,000 autonomous miles a day now, but it still has much further to go with its need for precise mapping and vast data.

Ali touches on the simplicity of the NAVYA system as well. NAVYA offers an autonomous, driverless, and electric shuttle that was providing rides at the summit. Here’s the NAVYA shuttle in Paris at one of the world’s largest airports:

Zach points out that the Tesla approach is based on deep learning, neural networks, and training the cars to respond like humans (with better reflex and more natural responses). This is in stark contrast to Waymo’s more closed-fenced approach. Waymo relies on geofencing, which has a predefined set of boundaries and programming limits. Ali expresses that data from both methods are valuable and necessary.

Ali notes that one of the differences for consumers as well as data collection is that Tesla chooses to make semi-autonomous features available to consumers to collect data faster. The problem is that average consumers are more vulnerable to collision risks than trained test drivers. The average consumer is going to be less tech savvy and may zone out or not follow instructions more frequently. Ali points to the infamous preventable crash of 2016. The crash could have been avoided if the driver had followed instructions. Since that time Tesla has implemented software updates that make it more difficult to ignore Autopilot warnings. There are big remaining questions regarding Tesla’s data collection approach that neither Zach nor Ali knew the answers to.

Discussion continues by covering the tricky matters regarding level 3 restrictions and when you can enable higher levels of autonomy. Research shows that even for trained professionals the mind loses attention too quickly when it doesn’t actually have to drive. Structural engineers have faced similar safety challenges, which is why long bridges are generally designed with curves and not as straight shots. The curves hold the driver’s attention better. So, whether developing bridges or semi-autonomous vehicle systems, we note that same task: How to make sure that drivers don’t check out too quickly. Easier discussed than done.

Ali notes that 5G internet connectivity is necessary for the highest levels of autonomy if the car is going somewhere it’s never been (if it isn’t operating in a geofenced area). There is far too much rapid processing needed otherwise. An autonomous car must collect an enormous amount of data, get it processed, and then send relevant commands back to the car’s drivetrain — all in the shortest time possible.

So, which will win the market — neural network or closed, geofenced approaches? Or will we end up with a mixture? It is hard to know since Ali highlights that the amount of data is still far too limited. Waymo is ahead in specific locations, especially in California, collecting data along familiar routes. Tesla, meanwhile, records everything every night and day that a driver does, but the depth, quality, and usefulness of such data is something hard for outsiders to make any assumptions about. One person once on the inside of Tesla’s Autopilot discussions is famous hacker George Hotz (aka geohot). For more on this topic, see his in-depth presentation and Q&A with Zachary: “Geohot: Tesla Autopilot = Apple iOS, Comma.ai = Android (CleanTechnica Exclusive).”

Related CleanTechnica videos and stories (a few of my favorites):

Save A Baby’s Life In Sierra Leone — For Real (#CleanTechnica Video)

CleanTechnica Chats With The Fully Charged Dudes (Video)

Tesla Has Strong Advantages In Race To Self-Driving Cars

Waymo Vs. Tesla: Who Is Winning The AV Tech War?

The Network Effect, Robotaxis, & Tesla

 
 
Follow CleanTechnica on Google News.
It will make you happy & help you live in peace for the rest of your life.

Home Efficiency




Tags: , , , , , , , , , , , , , , , , , , , , , , , , , ,


About the Author

Cynthia Shahan started writing by doing research as a social cultural and sometimes medical anthropology thinker. She studied and practiced both Waldorf education, and Montessori education. Eventually becoming an organic farmer, licensed AP, and mother of four unconditionally loving spirits, teachers, and environmentally conscious beings born with spiritual insights and ethics beyond this world. (She was able to advance more in this way led by her children.)



Back to Top ↑