- 1 day ago
A humanoid robot is a robot designed with a human-like body structure and capabilities, enabling it to operate in environments built for people ...An ultimate resource for the humanoid robot industry...
人形机器人是指模仿人类外形与行为特征、具备双足行走及类人操作能力的通用智能机器人。与传统工业机器人的“按代码编程执行”不同，人形机器人的核心目标是升级为能够自主识别、规划、作业和进化的智能体.
人形机器人是指模仿人类外形与行为特征、具备双足行走及类人操作能力的通用智能机器人。与传统工业机器人的“按代码编程执行”不同，人形机器人的核心目标是升级为能够自主识别、规划、作业和进化的智能体.
Category
📺
TVTranscript
00:20the robots are here it's not an invasion but a revolution these machines will soon be part
00:28of our lives and they look remarkably like us if we want to build robots we can interact with human
00:37beings then we need to behave like human beings robothespian is about recreating life with a
00:43machine walking is easy for a man but really hard for a robot when you see this robot you can
00:53see
00:53life inside it we've seen the future and this is what it looks like let's discover the best of
01:02europe's humanoid robots
01:36a new breed is evolving throughout europe a breed composed of creatures that look like us
01:43but they're not flesh and bone instead they are metal motors and silicon
01:50and very soon they may live amongst us these are the results of a fascinating line of research
01:56in robotics these are humanoid robots
02:05one of them looks endearingly like a child its name is i-cub created at the italian institute
02:12of technology in genoa it's now deployed in over 20 laboratories worldwide
02:22i-cub is used as an open platform for research and development in several aspects of humanoid robots
02:28from engineering to behavior thanks to its many sensors and its learning capabilities i-cub can
02:34interact with people respond to voice commands and recognize different objects i will teach you a new
02:43object great show me the new toy
02:49i will learn a new object
02:54where is the octopus
02:58i think this is the octopus
03:13i-cub is a complete platform it has sensors to see and to feel so it can feel touch and
03:20forces when
03:21interacts with the environment it has been designed with sophisticated hands which means a lot of joints
03:27that can move and the idea is that it can use the hands to explore the environment and what we
03:32do with
03:33the i-cub is to combine sensors and these sophisticated movement abilities to create intelligent behaviors
03:41these intelligent behaviors can allow the robot to interact with objects to interact with people
03:46and learn about the environment automatically
03:52and there's a very specific reason why this robot has been built to resemble a child of about three to
03:57four years of age
04:02building a child robot is interesting because it allows us to study the problem of learning
04:08so instead of building a robot that knows already how to do something we rather build a robot that
04:14interacts with the environment
04:22this baby robot is a perfect example of what roboticists throughout the world are striving for
04:28making humanoid robots more human
04:32it is a complex task that requires the perfect balance of engineering mechanics and cognitive studies
04:39some argue that it's also a bit like playing god
04:46the human race has always been fascinated by the idea of creating an artificial being
04:52as a companion for work and for social interaction creating something is a way for the human to transcend
05:03its mortality and creating something almost human is a very strong way of transcending the short limit
05:14of time that we have here on this earth
05:19the human race
05:19i-cub and other projects are laying the foundations for a future in which robots will share our everyday
05:25spaces and to do so they not only need to be shaped like us they need to interact with us
05:32in a safe way
05:33we need robots that we can trust
05:37bert a humanoid robot designed in the uk at the bristol robotics lab has been created with this goal
06:02and what we mean by this is to create a machine that interacts with a human in a shared space
06:09so that
06:09we don't have collisions or misunderstandings and one important concept in behavioral safety is shared attention
06:18so for example if humans two humans talk to each other and hand over an object several interesting
06:25things happen for example if i talk about an object i tend to look at this object just before
06:30and when i mention that object the other human then tends to look at the object too
06:36so in our work we try to create a robot that has the capability to do exactly the same
06:45per interagire meglio con le persone i robot devono essere sicuri robusti e attraenti
06:53meet ream a service robot designed by pal robotics in barcelona spain
06:59until just a few decades ago the word robot which comes from the czech robota meaning slave
07:05traditionally referred to large arm shaped machines used in factories for industrial production
07:11now the once derogatory word suggests something more like this
07:16but there are still many hurdles to overcome before we can safely integrate these robots in our world
07:21and not all these hurdles are technological
07:26le tre leggi della della robotica di asimo sono niente rispetto a quello che incontriamo ogni giorno
07:34a livello di certificazione per poter mettere i robot in entorni reali
07:40finora la legislazione ci obbligava a fermare il robot nel momento in cui si rilevava la presenza di persone vicino
07:49al robot
07:50mentre quello che un robot di servizio deve fare è giustamente andare incontro alle persone
07:56interagire con le persone e cercare di aiutarle nel modo in cui è stato programmato
08:16i robot come questo ci renderanno la vita più facile
08:21reem is one of the very first service robots available for purchase and it's been designed
08:26as a guide for museums shopping malls and other public spaces
08:32it can offer people information and guidance through its touchscreen
08:35and is able to interpret commands and respond to inquiries with speech
08:40i am part of your world now
08:44to make it all work a perfect integration between hardware and software is needed
08:48and real world tests can't start soon enough for robots like this
08:55when we meet someone we shake hands it's a basic form of human interaction that many people happily
09:01apply to robots as well but robotic arms and hands are often made of stiff mechanical parts
09:10the bird is still a very traditionally designed robot so it has stiff joints and stiff arms
09:18and it is intuitive to understand that a stiff robot can be potentially dangerous
09:24and really we humans are very different we are actually quite soft even though we have hard bones
09:29our muscles are soft they can move they are like springs they're not just gear boxes for metal
09:37that's why roboticists are coming up with a new model for the design of artificial limbs
09:43one that takes inspiration directly from the human body
09:49the next class of robotic arms may be inspired by bones muscles and tendons
09:58this robot is built from the inside out most humanoid robots just imitate the shape of a person
10:03but we're really trying to look inside and understand how the muscles have the tendons
10:07how the joints of the body all come together to produce the performance of a human body
10:29this is ecce robot and it looks eerily like a person with no skin and just one large singular eye
10:36ecce's design so intricately imitates the human anatomy that it even has to rest its back
10:44as part of a project we developed what we think is one of the most realistic spines ever made right
10:49down to the level of having gel in between the discs which turns out to be crucial for how it
10:53functions
10:55humans humans actually get shorter by about an inch during the course of the day as this gel
10:59compresses and the same is true of the robot which has to lie down overnight to allow it to stretch
11:03out again
11:13so we're dealing with the new ecosystems of robots supposed to operate inside the human body or
11:20outside the human body or close to the humans these robots have to be first of all very friendly that
11:26means they have to be flexible and have to be compliant in order to get compliant robots we need
11:33to develop new materials that are inspired by nature flexible materials responsive to electric stimuli
11:42and generally speaking materials that can make the robot as soft as a human being so this class of
11:49technologies will be developed together with the new robots so that we will limit the evolutionary
11:56pathway to reproduce beings like animaloids humanoids capable to operate across the humans in normal
12:03environments
12:14robots have a lot to learn from human anatomy especially in the case of arms and hands which
12:20are among the most complex components of a humanoid robot and the ones that could most frequently come
12:25into contact with people
12:31many researchers are creating new robotic limbs that will equip tomorrow's humanoid robots
12:40this is a new type of robotic arm it was inspired in the development by the human arm we wanted
12:47to build
12:47a full arm with a hand with which is capable to do the same motions and has the same strength
12:55and the same dexterity as the human has
12:58the arm is being developed at munich's german aerospace center and it's very close to the human equivalent in both
13:05weight and size
13:06even its peak energy expenditure is similar to that of a human arm
13:10just like our muscles this generates excess heat which requires the arm to be equipped with a liquid
13:16cooling system that looks remarkably like a network of veins and arteries and the hand at the extremity of
13:22this arm is among the most sophisticated ever built our robotic hand has 19 degrees of freedom which is
13:29approximately the same as a human does this is much more than a classic robotic hand which has approximately
13:35two to nine degrees of freedom thanks to this we can handle objects of different shapes very easily
13:41and we can grasp them and manipulate them like the human can do these robotic arms created at the
13:49industrial robotics and informatics institute of barcelona have a more traditional structure but an extremely
13:55innovative control algorithm they can come into direct contact with a person without causing any harm
14:03to put robotic arms in human environments they need to be safe there is a lot of investigations how we
14:10can
14:11make robots safe our approach is trying to design new control algorithms to accomplish that safety by
14:21construction not on the long-term control loop but on the very low level control loop and this means that
14:28we are investigating how we can make robots soft emotions that if they while they are performing its task
14:36they encounter a human they just not hit not hurt the human but behave in a very controllable manner where
14:45the human feels comfortable with him a structure that more closely resembles the human anatomy and has smart
14:52control algorithms is a good starting point to make these machines safer for human interaction
14:59but to be really useful in tasks and environments where humans normally work humanoid robots need to
15:05reach a proficient level of manual dexterity this is one of the research goals of aila a service robot
15:20being developed in bremen at the german research center for artificial intelligence the focus of our work is on
15:27manipulation we are especially interested in using two arms for manipulation and usually you require two
15:34arms when you want to carry heavy objects so you one arm can help the other arm let's say and
15:40the other is
15:41to do complex operation for example using tools hand tools drill machine and there we we need a certain
15:49dexterity of the hands we need certain flexibility we need sensors on the hands that can also sense
15:56and be very delicate with certain objects isla can be tele-operated from a distance by an engineer wearing
16:03an exoskeleton the robot faithfully reproduces every movement which makes it ideal for working in harsh and
16:10demanding environments like space in situations where it would be impractical or too risky to send a human
16:16being in this mock-up module of the international space station isla is working as a robot astronaut
16:24the arms only weigh five and a half kilos but they can lift weights of up to eight which means
16:29that the
16:29robot has a lightweight arm construction with a payload to weight ratio greater than one
16:35the hands are extremely human-like and offer an unprecedented level of dexterity and precision
16:46a slow delicate touch is not what we usually expect from a robot
16:50but in fact their computer brains and responsive motors mean they are in a strong position for split
16:56second reactions this robot is able to catch a flying ball or even tool
17:31back at munich's german aerospace center we meet justin one of the most advanced robotic platforms in
17:38the world justin is perfect for studying the interplay between perception and action
17:53the robot doesn't simply rely on its many sensors to produce a reaction but rather on a more
17:59complex system that imitates the way our brain thinks and works it's not hard for a robot to react
18:07quickly even insects can do this but to behave in a complex environment in a sensible way you need more
18:13human-like skills and this kind of cognitive capabilities we want to bring into this robotic
18:18platform this robot comes close to the agility and responsiveness of a six-year-old child
18:27this version of justin has a wheelbase but another variation called toro has been specifically designed
18:33for walking on two feet sometimes that comes easy for a man but not for a robot
18:41building a walking robot is challenging mainly because of the balance keeping the balance with
18:48only two legs is more challenging than keeping it for instance with four legs in contact with the ground
18:55on top of that we challenge our control algorithms by using smaller feet than usual
19:00our feet are only 19 centimeters long and this creates additional challenges for keeping the balance
19:07of course it's easier to keep the balance if you use really large feet but that makes the robot for
19:12the robot more difficult the task of going into environments where you have limited space for
19:17stepping for instance if you have very cluttered environments or very rough environments you might
19:23have very limited space for putting the robot on the ground
19:26you might think about walking you know even for us when we are kids and when we learn it it's
19:35it's quite
19:35tough and the same thing you know to do that with mechanics and robots is is a difficult thing in
19:42terms
19:42of coordination combining everything and so on and balancing at the same time it's also very important
19:47it's a very versatile way to move we're able to walk on a street or to climb on a rock
19:53in very difficult
19:54conditions something that the system with wheels cannot do and also probably developing robots
19:59with this kind of human way of walking can help them integrate better inside our personal environment
20:09barcelona's pal robotics have created multiple walking versions of its ream robot but as you can see
20:16there are still a few steps to take before we can see humanoid robots walk the way we do
20:38far camminare un robot humanoide è un compito difficile per due motivi uno appunto di vista
20:43tecnologico e l'altro è un problema teorico infatti ancora non abbiamo a disposizione dei motori che
20:49riescono a riprodurre alla perfezione l'efficienza dei muscoli umani e dal punto di vista teorico ancora
20:54non è stato trovato un algoritmo che riesca a risolvere il problema della camminata da un punto di
20:58vista diciamo completo e generale quindi stiamo facendo degli esperimenti che vanno via via migliorando
21:05l'efficienza e la robustezza della camminata
21:09iCup is also learning to walk progressing in its evolution just like a developing child
21:16and like a child the robot can interact with the environment to find its balance
21:24designing a robot to walk certainly a difficult task but what we more interested is rather how to make a
21:33robot that walks and does something useful I mean walking as I said is difficult but more difficult
21:42is to allow the robot to interact with the entire body which means using the hands while it's maintaining
21:47balance is what we're looking for in this respect what we like to do is to have a robot that
21:54goes
21:54around open doors open a drawer can grasp objects maybe it leans on a table and a piece of furniture
22:01if it needs to so to stabilize the weight but does something more than just walking I mean the
22:08research concentrated on walking per se but we think we have to do go one step beyond and
22:17manage to really do the interaction while the robot is walking
22:24so iCup was designed exactly to study development to study how you can build a robot which becomes
22:32progressively more and more skillful that starts by doing just simple movements like controlling the
22:40gaze of the eyes and the movement of the head and then you know progressively it becomes more and more
22:46skilled and try to grasp objects and and eventually at the hand the idea is to for the robot to
22:53be able
22:53to interact with humans through speech and gestures and any other way we usually use when we interact with
23:00each other but iCup doesn't only gain sensory experience of the world through walking it also does so
23:09thanks to its unique human-like surface skin the skin of iCup is a unique feature i don't think there
23:21are other robots in the world that can claim to have a full body skin that covers the robot is
23:26what
23:26allows the robot to feed forces and pressure when it's interacting with the environment when it's interacting
23:32with people the skin itself it's been like a touch screen on a cell phone but the technology has been
23:40adapted to cover the robot which means basically that it's soft which is unlike a touch screen that
23:49is rigid and flat the skin of the iCup is softer and curved to follow the body shape it's also
23:56sensitive
23:57to interaction normally with people but also with objects so normally a touch screen works only with
24:03our fingers in the case of the iCup it works with whatever the robot touches and as i said this
24:09is
24:10unique which means that allows the robot to really feel the environment to feel the interaction with
24:16people and this is a feature that we need because if we're interacting and we're collecting information
24:22then we can hope to learn something from this interaction if we cannot sense it
24:26then it will be very difficult walking touch and the ability to interact safely with the world
24:34are all crucial elements of a functional humanoid robot but something more might be needed to meet
24:40the ambitious goal that some robot makers are trying in cornwall england near penryn we meet robo
24:52thespian a robot actor that's already performed on stage alongside human actors
24:58it's designed as an entertainment machine but also as a research platform to investigate what elements
25:04are necessary to successfully humanize a robot
25:09when you see this robot we want you to see it as a person not some metal and and that's
25:16the aim of
25:16robot thespian is to move and to talk and perform like a human so that you believe that the metal
25:22is a
25:22human to be or not to be that is the question whether tis nobler in the mind to suffer the
25:33slings and
25:34arrows of outrageous fortune or to take arms against a sea of troubles
25:43at aldebaran robotics in paris we meet another robot who can move and perform with incredibly
25:48lifelike capabilities it's the smallest in our group and it's one of very few humanoid robots
25:55already publicly available its name is now now can be programmed to perform a vast range of different
26:05tasks and it's proof that the humanoid shape can be effective even at a smaller scale the humanoid
26:13shape that we have chosen for now is very important because we think that is the best way for the
26:18robot to interact naturally with people and if you want to create a real intuitive relationship with a
26:25robot this shape with a nice head with a little arms little eggs makes it very very sympathetic and
26:32people want to interact with it and based on this shape we have been working a lot on some expression
26:39behaviors gestures that make the robot living and and when you see the robot you don't see it as an
26:45object but as a little creature that has his own life inside and that makes the complete difference
26:51with a classical object like this lamp for instance
27:13these robots are already used in education in several countries there are 5000 robots in the education
27:20and research field to help young people first entering into this new world but maybe a world
27:27which is much more attractive people learning computer sciences students learning computer sciences
27:33with robots are more involved in their studies even if they are learning the right same things the
27:38very same thing but they are more involved in their studies because they are working on the future
27:42so so there are things that have been demonstrated which turns this robot a very very powerful way
27:49to attract young people towards science and technology to attract them and help them improving
27:54their skills in technology and simply to attract them into into science and it will help us having
28:01robots everywhere because this new generation of people who are learning with robots will bring the
28:08robots in our daily life but more than that there are some specific application in education we are working with
28:15uh autistic kids who are working with educators of autistic kids who are using the robot as a tool
28:22so educator is using the robot driving the robot programming or sorry using programs on the robot for having the
28:30robot teaching to autistic kids
28:32and autistic kids have uh other kind of way to interact with people and they are less comfortable in interacting
28:39with people and they are very good at interacting with technology
28:42then we can use this robot which is some kind of very rough uh human so they are able to
28:49interact with
28:50and thanks to that they are able to learn some skills of interacting with people
28:57there are now over 5 000 units of this robot in the hands of researchers and developers worldwide
29:03making it one of the most widespread open source robotic platforms in the world
29:13now has been designed as an open platform it means that it can be used by today by researchers
29:18and teachers and they can program the robot the way they want to use it and for us it's very
29:24interesting
29:24and very necessary because today we cannot provide many applications on it as you can have many many
29:31applications on your mobile phone on your smartphone today we don't have so many applications for now
29:36so it's important that the community we gather around now are able to develop their own application
29:42and with them we will discover what will be now used for and that is very important for us to
29:50bring
29:51not customers with us but a community of co-workers people able to develop with us the future of the
29:58humanoid robotics
30:00the open source philosophy also applies to icub even though the complexity of the robot and its high
30:06cost around 250 000 euros limits its availability to academic institutions and research centers
30:14but both the software and hardware of the small baby robot is being used as a collective research platform
30:21the aka project is open source uh which means that we give uh what we do both in software and
30:27hardware uh to other scientists uh so that we can share our results in turn other people can work on
30:34similar problems and give us uh the same results back to a common repository of results this allows for a
30:43faster and more efficient way of developing research because we can accumulate knowledge and this knowledge
30:49uh is uh you know available to everybody in a very open scientific way we because of this we now
30:58have 28 robots
31:00like this one built 25 of them are around the world mostly in europe but also in united states in
31:07japan and in korea
31:09um which means that icub in a sense is uh started as a small project but is now a worldwide
31:16uh community
31:19the ecce robot project is yet another example of collaborative robotics as well as being an open source
31:26platform it also employs innovative technologies for manufacturing and distributing its components
31:34we are committed to an open source approach to developing this technology partly because the
31:39technology is very challenging we need to involve a lot of people to deal with those issues and partly
31:44because the end user needs to be intimately involved with how the robot will function if this is to
31:49become an everyday part of our life our lives then we we need it to be something we're very comfortable
31:55with
31:55one technology that we're planning to use is 3d printing our aim is to produce a completely open
32:03platform the software the hardware and also the means to assemble it so we will use 3d printing as a
32:09way to distribute the designs in the hope that people will continue our work and then feed it back into
32:13the community
32:21open source started with software and we know that it produced really big revolution there but now we
32:27see more and more applied to the hardware that's also thanks to 3d printing or laser cutting the ability
32:32we have to produce pieces much much faster so we can collaborate remotely in doing things in the same
32:37thing and in making them better and that can really be a great asset when you think about robotic
32:43systems in order to speed up their evolution a complex discipline like robotics can only benefit
32:51from the open source approach allowing researchers from different fields to put their minds together
32:56to come up with new ideas and innovative solutions to existing problems for example we're really good
33:03at spotting every tiny detail that makes us human mostly involuntarily that's why they're trying to
33:09replicate some of these often unseen intricacies the most challenging aspect of designing a robot is
33:19everything they are very very complicated there's many levels you have to understand electronics mechanics
33:25computer programming you have to understand actually what people want to see it's a lot about
33:30managing people's expectations and then trying to make everything work at once so really i'd say
33:35the really challenging aspect is just management of very many tiny details
33:52if we want to build robots who can interact with human beings then they need to behave like human beings
34:06the concept of interface is very important for every technology and you know in the case of robots learning
34:12from the way we interface with each other as humans calling the human element then can become very very
34:18important rim nasce per aiutare le persone però per fare questo c'è bisogna anche che sia accettato
34:24le persone in questo modo abbiamo sviluppato dei movimenti involontari che robot riproduce in momenti di pausa per esempio
34:33il movimento delle mani il fatto di seguire lo sguardo delle persone o anche solamente camminando il fatto di
34:41imitare la camminata umana con il movimento del torso e delle braccia
34:47the small details are very important when it comes to designing robots human beings prefer
34:55interacting with beings that are similar to them the more similar we feel to something
35:01the more closer we feel to it so if the small details make a robot more human like it will
35:09facilitate
35:10the social interaction with a robot
35:13but no matter how much care goes into the details robots will always invite very strong responses from
35:20people in general we're attracted to them as a result of a mix of curiosity and all and their
35:26appearance can lead to wildly different reactions
35:33yeah people react in very different ways some people are quite afraid of a robot young children
35:38tend to just accept it as if it's another person we had one person who who threatened to kill the
35:45robot
35:46and tried to attack it that was when we were a show in Germany became very aggressive and said that
35:53the
35:54robot didn't have proper blood in its veins and he should kill it so people there are strong reactions
36:00and I think this is if you make a machine that looks like a person there's something quite profound
36:05about that it's almost the frankenstein story it's it's men playing god if you will recreating life
36:14mechanically and that upsets some people other people are very entertained by it
36:22something which came as a bit of a surprise was how much people liked the robot that was produced
36:27although it looks like a person with no skin and you can see all the inner details
36:31people weren't put off at all in fact they seem fascinated by it
36:40but what happens when robots really start to look like us at bristol robotics lab we find jewels it is
36:49a robotic head made with soft materials that can recreate amazingly realistic human expressions
36:54if it had a full body it would be called an android androids are humanoid robots that look like human
37:00beings even though they are largely confined to science fiction some have already been built mostly
37:06in korea and japan but creating an android poses more than a technological challenge it also forces
37:13roboticists to venture into a delicate realm known as the uncanny valley
37:23the uncanny valley is this idea that if you imagine a line where on the one end we have something
37:29completely non-human like like a ball with two eyes on it and on the other hand you have an
37:34actual
37:35human then as you move towards the human you enter a region when you get close to human but not
37:40quite
37:41where there is this sense of uncannous something that is just off that is kind of eerie and disturbing
37:46to people of course once you get to an actual human or something that is indistinguishable from a human
37:51that disappears but you have this valley before you get to that point or at least uh so the theory
37:56goes
37:56now i think it's more complicated than that um a lot depends on the context uh what your expectations
38:04are and even something very non-anthropomorphic like a disney cartoon figure if mickey mouse suddenly
38:09popped out there in the midst of it might actually be quite disturbing even if it would not even be
38:14close to human-like so i think we have to take this concept with a grain of salt there's very
38:22little
38:22psychological research about the uncanny valley so far it's a very hypothetical construct
38:30however i believe it's a very meaningful construct i've personally experienced the uncanny valley
38:38i was part of a project where i had a replica of my face made and put onto a robot
38:45and the replica was a
38:46tiny bit off and i felt mortified it was terrible it felt super awkward i ran out of the room
38:53i had a
38:54very personal emotional experience of the uncanny valley also the prosthetic hand i used to wear when
39:01i was a teenager had a cosmetic glove that was supposed to mimic the human hand but it was this
39:08tiny bit
39:08off and just looked awkward and awful so to that regard i've already had two very powerful emotional
39:15experience experiences of the uncanny valley so i believe it's real and yet the most fascinating
39:22challenge for roboticists lies not in recreating the perfect human appearance but the perfect human
39:28intelligence an intelligence that can learn and adapt while operating within the confines of a humanoid body
39:35one of the ways to reach this goal is through something called cognitive robotics
39:46so the key aspect of the investigation for the icab is what is called cognitive robotics which is aimed at
39:52studying what what may be called the mental aspects of of robotics not necessarily related to how you move
40:00motors or acquire images or other sensory but how you reason about information coming from your sensory channels
40:08and now you predict for example what is going to happen in in the future so the the it is
40:15a sort of
40:16an evolution of what once was called artificial intelligence with the main difference that it's not
40:23is not anymore a simple mental process like a chess chess player but it involves the movement of the body
40:31and the
40:32physical interaction with the world and with other persons with other agents
40:41robots in the future they have to be cognitive so they have to integrate millions of inputs from the
40:49uh sensorial system like vision hearing and so on and have to be capable to integrate and to interpret
40:57all this information uh ultimately to make decisions so neuroscience studying brain understanding brain
41:03cognition are again getting the most important sciences to make a robot capable of thinking
41:12to properly interact with a person the robot needs to understand a person's intentions
41:20so when we interact with each other uh in doing uh collaborative jobs for example like lifting objects
41:28together or uh closing a box uh together or passing objects one to the other uh what we usually exchange
41:37is not only forces but also information and we do that in sort of an implicit way so it's not
41:45uh and it's not a
41:46message which we uh which we uh state explicitly but it's the way we move the way we we we
41:52we reach for object the way we accelerate when
41:55uh we lift something this kind of of of messages are implicit uh in in the way we are built
42:04in our muscles in our
42:05body in our uh hands and the fact that the high cab is an anthropomorphic shape allows us to exploit
42:14these messages
42:15and try to understand what are the real features which are important to convey uh information when
42:21when we work together so this is one of the topics of of the research we are actually doing now
42:29is also able to learn by observation what is called imitation learning where we have a human
42:36operator showing some examples of certain tasks and the robot is looking at this person or getting the
42:43information from the movement of this person and after some examples the robot is also able to
42:48reproduce the same movements the same task like a child learns the coordination between hand and eye
43:00autonomously also with this robot we try to not pre-program it but to make it learn the task by
43:06simply
43:06executing it multiple times right now we're dealing with imitation and simple tasks but many scientists
43:16believe that in the near future we'll be able to create super intelligent autonomous entities and if that
43:22happens robot makers may be required to take ethics into account if a robot developed the same mental
43:33attributes as a human being it should also have the same rights as a human being not to do so
43:38would
43:39be a form of racism by the time a robot becomes capable of making decisions a question arises should we
43:55consider the robot as responsible for its actions i mean it sounds maybe not very mature at the stage right
44:02now
44:03but in the future this will be a issue a cognitive machine capable of making decisions might have
44:08responsibilities in case of mistakes should we punish the machine or not i mean something similar happens
44:17today with something much simpler like internet we underestimated in the past the importance of having
44:22extremely fast information anonymous information running on the net and now we're starting asking ourselves
44:30whether this has some headache implications if everything is well done so i think with machines
44:36that will be physically interacting with humans this will become much more important in the future
44:42and maybe it's time now to start thinking what could be a solution
44:53ethical issues should be part of robotic design because ethics come in where robots have to make
45:02autonomous decisions if a robot cannot process what it experiences based on prior knowledge it has to make
45:11a new autonomous decision and i'd prefer if some ethical guidelines would be in place for making such
45:19autonomous decisions intelligence is the most powerful force in the universe in this century we may create
45:26machine super intelligence machines that will surpass us in every intellectual dimension
45:33and if we ever build super intelligent machines how can we be sure that they will not be harmful
45:40or even a threat to our very existence
45:47it's a hugely difficult product to try to build a machine that is as intelligent as a human being
45:58it's also a hugely difficult challenge to figure out if you had such a machine
46:03how could you make sure that it would want to do the right thing that it would have some sort
46:08of human
46:08friendly values both of these challenges will ultimately have to be tackled but the sequence
46:16is crucial we've got to solve the second challenge the control problem before we figure out how to solve the
46:23intelligence problem we need to know how to control an intelligent machine before we build an intelligent
46:30machine being afraid of evil robots taking over the world is a natural thing
46:39it's caused by the way we perceive other beings the two core dimensions of social perceptions
46:47are warmth and competence if we perceive a robot we perceive a robot as socially cold
46:55but a robot is of course very competent it's a complex machine whenever we perceive something
47:01as cold and competent the reaction it instills in us is fear so a fearful reaction towards robots
47:09is very human in a way and therefore i think this is the explanation for why we believe that robots
47:17can be a threat i also believe that that is not warranted because robots being a threat implies them being
47:30intelligent and having evil intentions both of which i think is very improbable machines are evolving very
47:40very rapidly so the idea of machines taking over has been a recurring fear of humanity but i wouldn't be
47:47so concerned because i think that in between just a machine or a human if you have the human plus
47:54the machines will always win against both of the others if robots become a threat it won't be because
48:01of their guns it'll be because of their brains the most important question is not how we will treat our
48:07AI's or robots but how they will treat us in the long run our future will be in their hands
48:13so
48:29so