Topics
Latest
AI
Amazon
Image Credits:Getty Images
Apps
Biotech & Health
Climate
Image Credits:Getty Images
Cloud Computing
Commerce
Crypto
initiative
EVs
Fintech
Fundraising
Gadgets
game
Government & Policy
Hardware
layoff
Media & Entertainment
Meta
Microsoft
Privacy
Robotics
Security
Social
Space
startup
TikTok
transport
Venture
More from TechCrunch
effect
Startup Battlefield
StrictlyVC
Podcasts
Videos
Partner Content
TechCrunch Brand Studio
Crunchboard
get through Us
As robot videos become more viral, it’s important to develop a critical eye
In March 2008 , a roboticist in wintertime article of clothing give Big Dog a liberal kick for the camera . The buzz DARPA - funded golem stumbled , but promptly regained its foothold amid the white parking lot . “ PLEASE DO NOT kick back THE WALKING PROTOTYPE DEATH MECH , ” pleads the video ’s top commentary . “ IT WILL REMEMBER . ”
“ Creepy as hell , ” notes another . “ Imagine if you were assume a walk in the woods one day and saw that thing coming towards you . ” Gadget blogs and societal media account variously convulse out give-and-take like “ terrifying ” and “ robopocalypse , ” in those days before Black Mirror gave the world an even more direct stenography . Boston Dynamics had a hit . The video presently stands at 17 million views . It was the first of countless viral hits that persist in to this day .
It ’s hard to amplify the role such virality has play in Boston Dynamics ’ subsequent development into one of the world ’s most at once identifiable robotics society . Big Dog and its descendants like Spot and Atlas have been celebrated , demonize , parodied and even appeared in a Sam Adams beer ad . Along with developing some of the world ’s most in advance mechatronics , the Boston Dynamics squad have prove themselves to be super savvy marketers .
There ’s much to be said for the role such TV have played in spreading the creed of robotics .
It seems likely videos like this have inspired the career of unnumbered roboticists who are currently thriving in the field . It ’s a model countless subsequent startups have adopted to a all-inclusive range of success . Boston Dynamics certainly ca n’t be held creditworthy for any of those companies that might have taken a few shortcuts along the way .
In recent decades , viral robot television have grown from objects of curiosity among the technorati to newspaper headline - grabbing hitting filtered through TikTok and YouTube . As the potential reinforcement have increased , so too has the desire to soften the edge . Further elaborate topic is the state of CGI , which has become indistinguishable from reality for many viewers . Confirmation bias , attraction to novelty and a lack of technical expertise all diddle key function in our tendency to believe fake news and videos .
Join us at TechCrunch Sessions: AI
Exhibit at TechCrunch Sessions: AI
you could forgive the average TikTok watcher , for instance , for not understanding the intricacies of induction . Many roboticists have — perhaps unintentionally — add up fuel to that fire by implying that the organization we ’re seeing in picture are “ general purpose . ” Multi - aim , perhaps , but we ’re still some path off from robot that can execute any undertaking not hampered by computer hardware limitations .
More often than not , the videos you see are the merchandise of months or age of employment . Somewhere on a surd movement sits the hours of TV that did n’t make it into the last track , feature a robot stumbling , sputtering or break off scant . This is precisely why I ’ve encouraged companies to share some of these videos with the TechCrunch audience . Perhaps unsurprisingly , few have adopt me up on the offer . I suspect much of this comes down to how people comprehend such information . Among robotics , the hours and solar day of trial and loser are an indication of how severely you ’ve worked to get to the concluding product . Among the universal public , however , such golem failures may be see as a failure on the part of the roboticists themselves .
Back in a 2023 military issue of Actuator ( RIP ) , I praised Boston Dynamics for the“blooper reel”they published have Atlas lose its footing and falling in between successful parkour moves . As usual , a lot more terminate up on the cutting room floor than made the net excision . Even when not dealing with robots , that ’s just how things go .
A few weeks back , I attended a talk by theatre director Kelly Reichardt following a screening of her wonderful new(ish ) film,“Showing Up . ” She ingeminate that old W.C. Fields chestnut about never working with children or brute . In most case , I would likely tot innovative mechatronics to that list .
Along with CG / renders , creative editing is just one of many likely ways to dulcorate a robotics demo . More often than not , the intent is not malicious . A sentiment musicians frequently partake in with me onmy podcastis that once a song is bring out into the world , you no longer have control over it . To a certain extent , I trust the same can be true with picture . Choices are made to tighten things up and sweeten the presentation . These are an essential part of make consumable online TV . Especially in the eld of TikTok , however , circumstance is the first casualty .
There ’s no rulebook for what information one needs to let in in a robotics demonstration . The more I suppose about it , however , the more I believe there should be — at the very least — some well - defined guidelines . I am not a roboticist . I ’m just a nerd with a BA in originative piece of writing . I do , however , regularly speak with citizenry far smarter than myself about the subject .
Just ahead of CES , aLinkedIn postcaught my eye ( as well , it seems , the eye of much of the robotics community ) . It was penned by Brad Porter , theCollaborative Roboticsfounder and CEO who formerly steer Amazon ’s industrial robotics efforts . I seldom recommend LinkedIn pursue , but if you deal about the outer space at all , he ’s a good one .
In the piece of music , Porter mention that CES would likely be lousy with cool robotics demos ( it was ) , but total , “ there are also a lot of amazing conjuring trick - shot videos out there . class reality from stagecraft is hard . ” The administrator was n’t implying any of the damaging baggage that a word like “ stagecraft ” might have in this context of use . He was or else but suggest that viewers set about such video with a discerning and — perhaps — skeptical optic .
I ’ve been covering this distance for a act of years and have develop some of the acquirement to recognise machinelike kayfabe . But I still often tip on experts in the field like Porter when a demo feels off . Of course , not every witness has my experience or access to these folks . They can , however , fit out themselves with the knowledge of how such videos are sweeten — maliciously or otherwise .
Porter identify five unlike item . The first is “ stop - movement . ” This refers to a succession of rapid edits that make it appear as though the robot is go in path it ’s incapable of in real life .
“ If you see a robotics video with a batch of shape skips or tv camera stinger , [ be ] untrusting , ” he writes . “ You ’ll notice Boston Dynamics videos are often one snub with no tv camera cuts , that ’s impressive . ”
The second is simulation . This is , in exercise , the CG example I gave above . Simulation has become a foundational tool in automatonlike deployment . It allow people to run thousands of scenarios simultaneously in endorsement . Along with other computer graphics , automatonlike simulation has grown progressively photorealistic in late year . create and sharing a realistic simulation is n’t a problem in and of itself . The issue , rather , arises when you pass off such things as reality .
outcome three has a fun name . Wizard of Oz demos are call such due to the with child lifting being done by the [ individual ] behind the curtain ( pay no attention ) . Porter cites Stanford ’s Mobile ALOHA demo as an exercise . I strongly conceive there was no malice involved in the conclusion to go the ( still extremely telling ) demonstration via off - screen teleop . In fact , the “ robot operator , ” Tony Zhao , appears in both the TV and end credit entry .
Unfortunately , the coming into court come about two - and - a - one-half minutes into a three - and - a - half minute demo . These Clarence Day , however , we have to assume that :
Optimus fold up a shirtpic.twitter.com/3F5o3jVLq1
— Elon Musk ( @elonmusk)January 15 , 2024
For another instance that arrived shortly after Porter ’s post , take a facial expression at Elon Musk ’s X TV of theOptimus humanoid robot folding laundry . The video ran with the text “ Optimus close up a shirt . ” Eagle - eyed viewers such as myself spotted something interesting in the gloomy right - hand quoin : a gloved hand that from time to time popped partly into frame that play off the golem ’s bm .
“ frame in the Optimus laundry video just a few more inches to the leftfield and you would have missed what look like a tele - op hand moderate Tesla Bot , ” Inoted at the metre . “ Nothing wrong with tele - op , of trend It has some first-class program program , including training , troubleshooting and executing highly specialised tasks like surgery . But it ’s nice to know what we are ( and are not ) seeing . This strikes me as a obvious event of the original poster except key information , sympathize that his audiences / fans will fill in the gaps with what they conceive they ’re seeing based on their feelings about the messenger . ”
It could be unseasonable to accuse Musk of advisedly fully obfuscate the truth here . Twenty - three bit after the initial tweet , he add , “ Important note : Optimus can not yet do this autonomously , but surely will be able-bodied to do this in full autonomously and in an arbitrary surround ( wo n’t require a fixed tabular array with boxwood that has only one shirt ) . ”
As not - Mark Twainfamously noted , “ a Trygve Halvden Lie can travel midway around the universe while the truth is still pose on its shoes . ” A exchangeable principle can be applied to on-line video . The initial tweet is n’t exactly a prevarication , of course , but it can surely be class as an omission . It ’s the old paper thing of conceal your corrections on varlet A12 . Far more mass will be exposed to the initial error .
Again , I ’m not here to tell you whether or not that initial deletion was knowing ( if you pick out to apply the welfare of the doubt here , you may absolutely see the take after - up tweet as a genuine clarification of uncompleted setting ) . In this specific instance , I suspect most judgment on the issue will be directly correlate with one ’s personal flavour about its author .
Porter ’s next example is “ individual - undertaking Reinforcement Learning . ” you may do a deeper dive on support learninghere , but for the sake of brevity in a not - at - all brief article , let ’s just say it ’s a room to instruct robots to do job with repetitious substantial - world visitation and error .
“ Open a room access , pile a stoppage , turn a crank , ” writes Porter . “ larn these tasks is telling and they depend impressive and they are impressive . But a good RL engineer can make this work in a duet of calendar month . One step harder is to make it robust to different subtle variations . But generalizing to multiple similar task is very hard . In ordering to be able to tell if it can generalize , depend for multiple coach tasks . ”
Like teleop , there ’s absolutely nothing wrong with reinforcement learning . These are both invaluable tools for training and operating robot . You just need to disclose them as clear as possible .
Porter ’s terminal top is monitor environment and possible omissions . He adduce the then - recent video of Figure ’s humanoid lay down java . “ Fluid , single - cut , shows robustness to failure modality , ” he drop a line . “ Still just a exclusive undertaking , so claims of robotic ’s ChatGPT bit are n’t in evidence here . output quality is great . But you ’ll notice the robot does n’t lift anything heavier than a Keurig cup . pick up mugs has been done , but they do n’t show that . Maybe the robot does n’t have that forcefulness ? ”
When I speak with Porter about the intricacies of the post today , he was once again agile to point out that these notice do n’t detract from what is authentically impressive technology . The issue , however , is that our brain have the tendency to fill in disruption . We anthropomorphise or humanise robots and assume they learn the way we do , when in world , watching a robot unresolved one door absolutely does n’t secure that it can open up another — or even the same door under dissimilar firing . TVs and movies have also given us unrealistic expectations of what robots can — and ca n’t — do in 2024 .
One last point that did n’t make it into the post is speed . The engineering can be painfully slow at times , so it ’s common to cannonball along things up . For the most part , university and other enquiry facilities do a good line notice this via a school text overlay . This is the way to do it . supply the apt info on screen in a mode that is unmanageable for a click - thirsty influencer to pasture out . In fact , this phenomenon is how 1X got its name .
A late video from the company showcasing its employment of neuronal networks draws attending to this fact . “ This TV control no teleoperation , no computer graphics , no gash , no video recording speedup , no scripted trajectory playback , ” the caller explains . “ It ’s all controlled via neural networks . ” The result is a three - minute video that can finger almost painfully tiresome equate to other humanoid demos .
As with the blooper videos , I clap this — and any — sort of transparency . For sincerely slowly moving robots , there ’s nothing incorrect with speeding things up , so long as you stick to three import rules :
Much like the songwriter , fellowship have to acknowledge that you ca n’t ensure what happens to a TV once it belongs to the world . But ask yourself : Did I do everything within my power to stem the spread of potential fakery ?
