Here are the timestamps. Please check out our sponsors to support this podcast. 0:00 - Introduction & sponsor mentions: - LMNT: drinkLMNT.com/lex to get free sample pack - Athletic Greens: athleticgreens.com/lex and use code LEX to get 1 month of fish oil - ROKA: roka.com/ and use code LEX to get 20% off your first order - Indeed: indeed.com/lex to get $75 credit - BetterHelp: betterhelp.com/lex to get 10% off 1:08 - Robots in science fiction 6:49 - Cozmo 32:04 - AI companions 38:59 - Anki 1:04:33 - Waymo Via 1:36:10 - Sensor suites for long haul trucking 1:46:06 - Machine learning 2:04:03 - Waymo vs Tesla 2:14:38 - Safety and risk management 2:23:42 - Societal effects of automation 2:34:47 - Amazon Astro 2:39:12 - Challenges of the robotics industry 2:43:39 - Humanoid robotics 2:50:42 - Advice for getting a PhD in robotics 2:58:13 - Advice for robotics startups 3:09:19 - Advice for students
Talk less, listen more. Your guests are saying far more intreresting things than you. Please talk less, you're like a piece of chalk screeching on a table. I would guess that you talk for at least 40% of the podcast, at least that's how it feels for me.
Mr Lex , please have guests on that use math and science to prove or disprove the existance of luck and intangibles . Please incude sports , gambling and surviving lightning strikes / airplane accidents ect .
3 Hours 14 Minutes and 13 Seconds. Why could you not have made this 2 seconds longer? Do you hate pi?! (all kidding aside thank you for the great interviews. Yours is my favorite podcast)
Robotics is such a refreshing technological endeavor because it recognizes the inherent beauty and reality of physical space, contrasted with something like the metaverse that sees material reality as an obstacle to experience.
I agree, assuming it wouldn't get hostile and focus entirely on the engineering issues and not accusing/assuming something about the future (if it'll work or not).
It doesn’t matter what happens I like you. You are bringing all the scholars to popular domain. We need them to come out to talk to the public more often.
The problem with Waymo is that they are not selling anything(unless you live in a tiny square in phoenix), not selling cars, not selling their drivers, people wants self-driving today will just go by a Tesla or a comma three. At end of the day, this is a business, Waymo's business model does not even exist.
I love how lex looked at his hedgehog . At this point, we all know it’s an AI ML that observes all of his talks and actually becomes the new internet. P.s - i’d love to help you work on that
Lex, I'm familiar with you from Joe Rogan's podcast but only recently started listening to your podcast. I like the kind of structured/kind of freeform nature of the conversations with intelligent and thoughtful guests. I'm enjoying it very much. Please keep going. Thank you.
As an automotive engineer with ADAS experience this was a particularly interesting discussion. I'm only halfway through but I don't know if a Waymo senior engineer will really address some of the biggest issues in localisation, perception and planning that I have come across in my work. I think there are also so many practical issues with things like Waymo Via that the robotics experts just don't get. From my limited understanding, half the job of a truck driver is keeping the truck running well as well as loading and monitoring the load. A self driving truck can't pull over and fix a shifted load, replace a busted tire or popped off airline?
this is so true. The best drivers can do on the fly mechanical troubleshooting in order to get the load in on time. I guess they could monitor all the systems and take trucks out of service before there's a breakdown, but there's really no substitute to having someone on the spot.
Probably autonomous trucking will be trains/convoys of multiple trucks, maybe for every 20 trucks you have a human manned service vehicle. But you make a valid point, the actual considerations of practical use of autonomous vehicles for trucking
@@hankkingsley9183 If you have a human riding along, why wouldn't they just be in the front truck and could drive or "lead" the others. You could even link them together and run them on metal rails separate to the road and this would increase safety - but then it would just be a train :) I used to work on this stuff and love the technical problems of automated/semi-automated cars. One of the reasons I walked away was the marketing department. Another was the constant expectation of "it will be here in 2-3 years" when realistically I can't see a level 5 vehicle being equivalent to a human car in less than 2-3 decades. No one had the guts or the ability to explain it to management who were competing with other companies making the same promises.
@@aless5207the train idea is actually interesting lol. By the time the amount of time and money has gone into R&D for making AI systems sufficiently good at replicating the role of a human trucker, why not just redesign the road infrastructure completely into something different with AI. One day maybe. But if it's going to take 40 years anyway
When I facilitate bamboo bicycle making workshops often the "cheaper and fun" thing comes into play as people realize how using a natural material (which is already engineered for you) quickly allows one to be creative and get "good enough" results fast to make a fairly large impact on how they actually experience personal mobility - without having to get stuck in the bicycle industry hype and costs. So, yes, I agree perfection is often not fun and just being good enough can offer faster more creative experiences across the board. In the case of the bamboo it is a robust material but with very simple ways to make it useful so it's a really good example of that balance between practical results and creativity.
The conversations in your podcasts are always so fluent, intelligent, interesting and inspiring. As a non-science person, I truly enjoy all your episodes and love all your guests, (and even more when they quote or mention the books that I’ve read). The dialogues about life and advices for young people are so valuable. The passion, the vision, the hardworking, the sincerity, I don’t feel so lonely any more.
I appreciate the parts of himself that Mr. Fridman allows to be seen. My admiration and respect happened after 3 seconds but has grown so much and in connection with that his handsomeness as well. I wish him all of the beauty in the world. Having said that, now I will have to forget all about him
I pre-ordered a Cozmo as soon as I learned of it back in Jun 2016 and received it in October. It's the best physical toy I've ever bought. Even though it's always in the back of my mind that I don't have enough time to play with it (with my kids or by myself with the SDK). I was also super-sad to learn about the end of Anki.... At least Cozmo has been revived by Digital Dream Labs, albeit at a much higher price.
The problem with Cozmo for most families was, do you want video game money, or do you want a Cozmo? Do you want a new tablet, or do you want Cozmo? Tech. is nickel and diming the average American household with cellphone expenses, home internet cost, home cable entertainment cost and the litany of membership services within all of it. If they were fifty bucks, they would have flown off the shelves.
Issiac Asimov Mule Character in Foundation Trilogy strapped a Helmet on and commanded the Universe to die. I asked Issiac about this in 1973. He had no comment. If a Squirrel runs in front of your Dump Truck while you're hauling 12 Tons of Asphalt at 60 Miles an hour around a curve in Wisconson. What's the right Choice?. I got a Pancake Fisher around Lake Superior once too. What's the right thing to do? This happened to me, Pancake Squirrel, Pancake Fisher was my decision. I'd love to see how AI handles the Squirrel/Fisher Pancake moral dilemma within the context of rank ordered Choices in Truck Driving Decisions. Great show! If you can get Wolfram again, I would love to see more on Patterns and Cellular Automata. Great guest! Many don't know Issiac made a great series of Science Books for Grade School Children. Issiac was very happy I remembered. We need to improve Science Education in Grade School. Odd Times
So many valuable insights - I mentor high tech startups (having done it on my own) - I will suggest they watch this discussion! On another note - can you release a “where we are now” video regarding AV tech? I drive a Tesla and am fascinated by this tech -
Cosmo and the Anki brand was my son’s first real connection to robotics, electronics and code. He had access to these things before, but the accessibility and usability, the anthropogenic nature of Cosmo, really got him hooked. We were so sad to hear about the Anki news.
Hey Lex, is there any way to have Daniel Tammet on your podcast? Would be outstanding to get a deeper insight into his incredibly rare gift of arithmetics and language. Brilliant podcast as usual!
Great is always. By the way, is there any chance for podcast with Christopher Ulmer, Special Books for Kids youtube channel? He is a great human being.
I love you and your podcast Lex! But as a truck driver for 23 years, I have to say the latest technologies I really hate! I miss the days of manual control, hate the self braking which may be safe at times but more dangerous if the roads are bad. Also, it's just more crap to go wrong and cost a lot to fix. How about making better ABS, and sensors that actually work and last in the harsh weather conditions before rushing to self driving trucks to put millions out of work. I could go on for hours about how unsafe the new technology is in certain issues, but I won't bore you. Keep up the great podcasts, but this one really frustrated me because of all the things that still need to be better before pushing more technology that is going to put people on the roads in danger.
Lex, at around 27:00 you talk about characters acknowledging that you exist as a player. As a curiosity look at Doki Doki Literature Club. I know it sounds really weeb, but it is a really cool game.
-People constraint themselves a lot when they try to replicate a robot in a human form. It has many downsides. First the cost. Second that the articulation of human body is so complex that it takes a gaint amount of joints to actually develop it. - Less constraints makes it difficult to express emotions. Graphic Designers usually challenge themselves by restricting themselves to just balls and they have to express their emotions through them only. - We get a lot of amazing innovation in movies and games space. But the gap in betweeen is left out i.e. the toy space. - Things that feel like really constrained in the physical world, they have an amplified impact on people where the exact same experience virtually would not have anyone near the impact but seeing it physically. Basically bringing things physically totally changes the experience and has a lot larger impact on the people. - Perfection should not be the aim when you are exploring things like robotics. Even humans start imperfect then go onto becoming perfect. Start from 15:20
LOve the podcast Thanks When creating your robot why not record the whole process? No editing just raw screen capture as you go through the process. You would get thousands of people signing up on Patreon to watch as premium access. Help found your startup without VC cash
The biggest value proposition for a humanoid form robot like SpaceX's is tele-presence, particularly in circumstances where it is inconvenient or dangerous for a human to work. MARS for example... However, even there I think SpaceX will find that they get nearly all the value of the humanoid form for telework from mounting a humanoid torso, head and arms on something that looks a lot like the lift bucket vehicles SpaceX commonly uses for working on its rockets.
The trolley problem has a critical flaw in which humans assumes level of kill either way seems likely. In practise, that never happens. If the truck does Hard Right or Left and fully On-the-brakes, it might hit some animals nearby. Like crickets in the grass..... And None of the humans. Also, if the gas tank suddenly rupture upon impact, it could kill many more people than you think!.
watching this conversation now its clear that no company would be like anki for a long time. especially ddl whose owned the rights of anki's product right now. bummer. RIP cozmo and vector.
My favourite podcast. Question - shouldn’t autonomic tracks have THE SYSTEM first? I can’t imagine that some of the cars are autonomic and some are not. They should be a part of one smooth system to work well. Question 2 - where i could submit the design for Apple car? There should be a competition for the shape. Some of the projects I’ve seen are disgusting as MULTIPLA car…
My husband Kurt is watching with me and right on cue... his answer was WALL-E...and R2D2 at the same moment. It was eerie to be honest. I guess self proclaimed "nerds" think a lot alike. Lol
for waymo the problem i see is that if they are successful at a large scale in the future we the public don't get the same kind of benefits we would get if tesla is successful to level 4. that is, waymo's approach is very expensive to maintain and so won't be cheaper or much cheaper than employing people to drive cars. so in many scenarios there's a feeling of 'what's the point' to it.
If eventually most cars on the road get to level 3 pretty much everywhere, might the risks of driving go so far down that all those cars effectively become level 5? I.e all the odd circumstances where another car does something unexpected simply cease because no humans are driving any more.
But many limbs rotate in place. Like the wrist, fore-arm, shoulders etc. lots of limbs simply move vertically in space like the shoulders. Or am I misunderstanding?
@@ekothesilent9456 well, hold up your hand, bend it left or right, forward or backwards and watching the most extreme part of your limp, (finger etc) you'll see it moves at an arc,
The perspective I got from this when it comes to the Cosmo toy it sounds like a very elitist conversation. I don't know why they keep referring here $200 for a tiny hand sized toy that makes cute noises and stacks boxes "cheap". Maybe if you got it like that but most normal folks don't and that's bot smart for mass appeal and sales. Even though he was warned it would be seasonal he seemed shocked that it turned out to be seasonal. He may have come up with an awesome robotics creation but there was no further thought into practicality. It sucks his company tanked but it was inevitable. Lex is a great host and keeps the conversation flowing. His love of robotics shines through.
Throughly enjoy your conversations with such diverse, eminent and interesting people. However, I’ve noticed that they are predominantly men and as a woman would love some more female content 😀 Leslie Udwin is an extremely interesting and amazing woman I’ve recently discovered, she’s out there trying to create major change across the world and should be more recognised and acknowledge for her work. She is a human right campaigner, a BAFTA award winning filmmaker and 2019 was awarded the UN Women for Peace Activism award. I think she be a great guest for the Lex Fridman podcast. Thank you for bringing us these great conversations 🙏
OK, Lex: time to build the RoboPet. I wish an enhanced Cosmo were available, suitable for adults as well as kids. Cosmo was seasonal, but pets are not. I think a RoboPet could capture much of the pet market that is today worth billions.
I liked your comment as soon as I read it, but then I thought about it, it would solve lots of problems with pet food production,vet bills ,cruelty and the worry about leaving them alone to wreck your house. Then I thought about robo guard dog 2.0 with built in Tommy gun or something lol
@ConecoTheAwesome I am thinking: soft, furry, animal like, but not an imitation cat/dog. It comes when you call it, exhibits curiosity/independent behavior, likes to be stroked, and perhaps has some verbal skills. People can bond to anything (cars, teddy bears, robo-vacuum cleaners, and lizards). A little effort to make the robopet more companionable than a fish in a tank would make it a viable pet replacement. It would sell millions and save their owners a lot of money (over 'real' pet). Vets will have to learn new skills 🙂
American engineer Keith Tantlinger's inventions played a major role in containerization. A steel box. Standardization trivializes autonomy. I know Kenworth is running EVs at LA port. I bet Freightliner is running AVs, no one wants to be a longshoreman anymore because of automation, and that's why there's 100 ships anchored off LA port. This dude could literally be responsible for a global supply chain shortage and this may very well go down as the single biggest disruption of an industry of all time...not the tech itself but its spectre. The disruptive aura that precedes it.
It's a tricky idea asking negative questions as he won't want to break the flow of the conversation, you're right though, I would of liked him to be asked about the responsibility of creating something powered by the cloud knowing people will want to use their devices long after it's economically sensible for you as a company to run the servers, especially if the end-device is a child's friend
Waymo's solutions strike me as poorly engineered. The car sensor pod is huge because it needs the training wheels of lidar. It changes the drag coefficient of the vehicle, and probably a lot of weight. The semi truck sensors hanging off the side...... what happens when one gets smacked by a bird which often happens to side mounted mirrors? They are also massive and add more drag as well. Seems like a very clunky solution, both for cars and semis
