Humanoid Robots - Science topic

Insecure communications, authentication issues, weak default configuration, denial of service attacks, risk and vulnerabilities, and privacy issues have raised several security concerns in robotics. In consequence, the demand for cyber security solutions and services in robotics is increasing to prevent robot from being hacked, monitor the vulnerabilities, prevent the data from being tampered, and to prevent unauthorized access...

Artificial Intelligence and Machine Learning are being applied across major industries and applications, including the cyber security in robotics. These technologies are used to detect the cyber threats based on analyzing data and identifying a threat before it exploits a vulnerability in the information security in robotics. Moreover, Machine Learning remains a key focus area of companies operating in the cyber security in robotics market, as it enables the robot to predict threat and observe behavior anomalies with more accuracy. Machine learning and artificial intelligence are also used for detecting and tracking the active phishing sources...

Muhammad Yeza Baihaqi

Unfortunately, there is no cheap alternative, Nonetheless, Promobot claims that it may be beneficial in both households and businesses. The robot costs between $20,000 and $50,000, depending on choices and customization. Join CNBC at the @ Work: People + Machines Summit in San Francisco on November 14 for more on technology, change, and the future of work.

combination of V-Rep and MATLAB will be perfect

In my university, they used to draw using shape tools in PowerPoint, very easy. It is only a stick diagram.

Muhammad Yeza Baihaqi

If you want to study robotics, you need first master computer science, coding, physics, and linear algebra. From there, you may take robotics lessons, join clubs and compete in competitions, and improve your fundamental robotics skills over time.

However, the field of robots is as complex as it can be. However, if you have a strong desire to conquer the unfamiliar and the complicated, mastering robotics should be a breeze even in your mature years. Some of the reasons why you should start learning robotics right immediately are listed below.

Dear Baydaa A. Hassan,

I also wish you good luck and invite you to our discussion.

Thank you, Regards,

Dariusz Prokopowicz

Thomas Wayburn,

Like I said above, human intelligence, from his/her capacitation created smart technical taskmaster technology, there is not artificial intelligent, it's the product of by real human intelligent.

I call it misuse of phrase.

Regards,

Fatema Miah

As was clearly described by Eugene Veniaminovich Lutsenko, material world is only part of the universe. Not the whole. Science is studying material things. The rest is expected to be non-existent.

The fact is that there is a piling evidence on the existence of the after-life or before-life where energies that we call spirits resides. When spirits enter the material world, they get accompanied by atoms.

Spirit can live without a body a.k.a. matter. Matter cannot become alive without a spirit that is making it alive.

Here we come to the important point: "What we spirits/souls?" Those are quantum fields that we are unable to detect by the current technology. Spirits are the oscillating specks within all encompassing field of the whole Universe.

After-life is a serious research subject that is working usually with deep hypnosis with regression of subjects into their past lives. Then researchers go and check all details and in the most cases find relatives of deceased who lived exactly same life as described by a person in regression. Such research was started by Dorothy Carron. There are followers like Newman and co-workers.

Dear Fatema Miah,

Yes, you gave another example of the perfect application of robots, i.e. in difficult conditions for human work. Of course, robotics must be developed in a variety of applications, but only under full human control.

Thank you, Best wishes,

Dariusz Prokopowicz

Hi Jonas,

I'm not quite familiar with the Unity-specific limitations, but have you considered controlling the robot from "outside" using a software-in-the-loop approach with a virtual robot controller or robot controller simulation module? Basically forget about modeling the kinematics (and dynamics etc.) in Unity, but mirror the internal robot model state from the VRC. We used that for simulation bin picking processes in Siemens NX MCD, also with physics enabled, and it worked just fine . Also, you get a much more realistic robot behavior, since it is controlled by the actual controller and program. However, you will not get a physical feedback to the real controller (joint torque etc.). If you want to imitate a collision detection through the robot for HRI topics, I would either add simulated force sensors to the robot and tool (if Unity supports that), or just use a collision-based setting of a variable/signal that triggers as a safety stop on the robot. See also here:

Greetings

Max

In any means, we must not forget that AI is directly associated with wars and it is originally launched for the destruction of humanity.

Dear Thomas Anthony Troszak, Thank you very much for the rich, substantive information and explanations on the topic artificial intelligence. Thank you for the information provided and the link to interesting data in the above-mentioned issues. Very interesting considerations. I am very glad that the discussion on this important topic is developing.

Thank you very much and best regards, Have a nice day,

Dariusz Prokopowicz

...To increase the autonomy of humanoid robots, the visual perception must support the efficient collection and interpretation of visual scene cues by providing task-dependent information.... Grotz, M., Habra, T., Ronsse, R., & Asfour, T. (2017, September). Autonomous view selection and gaze stabilization for humanoid robots. In 2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) (pp. 1427-1434). IEEE.

The server you're using still may collect some logs, it's not a big deal but some notice could be needed.

I am mathematician and love AI (Mathematics & AI is as body and soul), mainly during my PhD in AI which was based on mathematical/statistical modeling for applications in computer vision. During my PhD training, my main concern was, on how to utilize and bridge AI with mathematical/statistical modeling for automatic patter recognition. To me it is really the ever thrilling fun in scientific world. For a mathematician when working in AI, obtaining fast computations and visualization with instant observations by experiments is really a fun-play.

I do not see any drawbacks of AI. It is a problem solver technique in faster and optimal way, so drawback is no point either. Concerning to its pros & cons, I would like to say, it is by the user well, how they use it?

check this link maybe help to the discussion

https://pdfs.semanticscholar.org/0d8a/22f3981375bccd8c4664ac3d58f2bf6dc1d1.pdf

Mayuresh Amberkar okay something a bit ambitious but actually it was something I wanted to do few years ago but I gave up because not enough time and the technology was not there.

You can consider a CNN which is detecting the face expression/emotion (or some body attitude in case of elders) and produce either:

- Motion of the Nao body either mimicking the emotion that was detected or something different in order to change the mood of person

- Synchronize the body motion with a specific speech - try to change the tone if you can because it is an important factor in the voice that can influence the mood of the receptor.

You have an full loop between perception and action (kinematics) and an interesting area of research at the intersection of psychology and robotics.

Cheers

Accelerometer alone should be more than enough for such task as it has been used widely in activity recognition studies. If I were you, I would start with the accelerometer only and collect data from experiments to what you consider "smooth walking" for your robot. You could also try extracting time/ frequency features from the accelerometer data and use that to compare between desirable vs undesirable walking.

There is a very good book, published in 2018, and named "Hello World: How to be Human in the Age of the Machine", It might help.

Hi,

The latest studies in any subject are indexed in Google Scholar. The next best thing is to search a relevant database in your research area. You may try locating new material on ArXiv, which is a preprint database for physics , mathematics , computer science and other sciences. This repository is free on the net and you could locate there material before it is published in a formal journal.

Most of the applications I see are educational. In addition, it has applications in medicine, therapies, among others.

I recommend checking the papers of the IEEE-RAS International Conference on Humanoid Robotics (Humanoids).

Dear Tahir,

I suggest you to see links in topic.

Human–Robot Interaction: A Survey - CNRS

-Socially intelligent robots: dimensions of human–robot interaction

-Robots Working with Humans or Humans Working with Robots? - MDPI

-Human–Robot Interaction: Status and Challenges ... - SAGE Journals

-Computational Human-Robot Interaction - Now Publishers

-Human-Robot Interaction | The Encyclopedia of Human-Computer ...

-Conference program - Human-Robot Interaction

-Journal Club — Social Robots

www.so-bots.com/journal-club/

-HRI - ``In the wild'' In Rural India - Social Robots in the Wild

Best regards

Dear Eduard,

I suggest you to see links in topic.

-Innovation Potentials and Pathways Merging AI, CPS, and IoT - MDPI

-A blockchain future to Internet of Things security: A position paper ...

https://www.sciencedirect.com/science/article/.../S23528648173029...

-The Internet of Robotic Things: A review of concept ... - SAGE Journals

-Industry 4.0: the fourth industrial revolution - guide to Industrie 4.0

-Mobile Cyber Physical Systems: Current Challenges ... - IEEE Xplore

-Prospects of robotics in food industry - Scielo.br

-The Future Internet of Things and Security of its Control Systems - arXiv

-Robotics and industrial automation enabled by 5G - Ericsson

https://www.ericsson.com/.../industrial-automation-enabled-by-rob...

Best regards

If you are asking in the swing phase of gait I like that recommended article "Sagittal Gait Synthesis for a Five-Link Biped Robot ". If it is during the stance phase of gait this is variable based on many factors: Gastroc tightness, pes planus vs pes cavus, and joint positions of the extremity above the ankle in the extremity.

Looking at what some popular humanoid robots could do, we can see the progress in this field is quite impressive. I supposed Sophia (social robot) can be used as a benchmark in this case (robotic science).. and she even obtained her citizenship lately

The design of humanoid robots involves significant challenges which include the scope for self recharge, automatic restoration to adjust to varied climatic conditions and serious issues related to control aspects.

Have you noticed being somewhat dependent or attached to your domestic robot? It is truly smart? Can you use it as your phone, TV, Health Support, Car and Garage controller, Gardener, plus....? Or it is just a simple Robot that performs very specific tasks?

Dear Artur, thank you. Indeed, so many questions and so many excellent answers. At this point in my career, I take interest in things that I did not have time to investigate or study in depth. There was always so many other other thing to study, discover and understand. Somehow I wish I could be very young again so that I can start learning so many wonderful and fascinating subjects. Mind, I did learn little bit of sixteen foreign languages , and did all complete all degrees possible, and yet, I am somewhat disappointed that I missed so many other things.

Dear Kirk, I agree the Human Spirit is far beyond machines and always will be. It is people who may choices based on their own set of values and believes. Machine can only do what they were programmed to do and indeed not reason like human. The future surely will bring lot of advancements in AI, Robotics, Automation, Bio-technologies, plus much more, but will be it be enough to conquer Human Spirit, I found it difficult to believe.

I would also request the same. If its possible, I would to like to explore goals of this project together and collaborate 

Thanks

Sphinx is a flexible framework for research in speech recognition. use that

Thanks Mr. Avisek. Yes, I have a connection with them, they supplied me P-E loop tracer. Thank you very much once again.

This project is an engineer's dream. It has elements that require the project team members to creatively blend established (even old) techniques with new ones, in "outside the box" ways. Teleoperation in real-time is not possible, so giving detailed directions to the onsite exploration vehicle won't work. So, develop primitive routines that can be strung together like beads using higher-level commands. The rover should understand general commands from mission control, like "explore quadrant G31 and report", leaving details of how to accomplish that task to itself. An expert system would be an excellent starting point, with the "experts" being the development engineers and scientists. The rover sensors would allow monitoring planning and execution. Any exceptions to a routine, such as bumping into a rock, or approaching a precipice, would be reflected to control on Earth, to facilitate improvement to existing routines and addition of new ones. Then, creating a "curiosity" attribute to blend a variety of exploration methods would be helpful. How would a robot look for interesting planetary features and objects, or avoid weather? Imitating human behavior is a great place to start: nature has provided the plan. I gave a 3-hour presentation at MIT on this approach 7 years ago: “Tutorial: Biologically-Inspired Object-Oriented Design & Object-Oriented Programming for Robotics and Artificial Intelligence”, 15th IASTED Intl Conf on Robotics & Applic, MIT, Cambridge MA, Nov 1-3, 2010. I've given 26 presentations at NASA Houston since 2006, most involving artificial intelligence and robotics for space applications.

Hi Nour,

Generally the velocity of the robot trank is not constant during the walking motion. The velocity changes depending on the Trank position with respect to the position of the standing foot.

You can determine the average velocity simply by measuring the step length and divide it by the time needed to take the step.

Hello technion,

Indeed, I need to see and experiment with other methods and tools of work. I will be glad to see what you have done as part of your research.

Thanks in advance

Maurice

i did colleborative research with dr. akihisa, ninon university

Microwave or mm-wave would be perfectly good - broadband best.  It is not speed that matters but wavelength.  Look up monopulse accuracy, for instance.  You will need a spread of wavelengths or a spread of antennas to get accurate direction in a room.  You will need some kind of range discrimination otherwise you also locate the directions of the reflections of things in walls,  like being in a hall of mirrors.  Range discrimination requires either a spread of wavelength or widely separated antennas or a bit of both.

Thesis

Hi Anil: 

You can consider this problem as a kind of 2D motion planning. Assuming the lenght and the turning you can perform with each step known (which will be different for each model of humanoid robot), you can model your problem as a graph where the states are the poses of your feet and the arcs the connectivity between these states (i.e. if you are in a certain position you can advance an step, turn right or turn left, which leads your robot to have a new pose for each one of the feet). You can use the information of the robot state to check possible collisions in the known map. This way to model the problem creates a graph in which you can search using a discrete search algorithm like A*, ARA* or AD* (if you are interested in replanning too). 

I've used this approach successfully in wheeled robots and UAV's, and it would be easily applicable to navigate with humanoid robots too. If you take a look to the publication I attach you maybe you can find some useful information about this approach Sections III to V. Even if in this case it is applied to a wheeled robot you can adapt it easily for a humanoid.

I hope my answer helps you.

... That's an interesting idea. With AI, we are so busy trying to make algorithms perfect. But to emulate human intelligence I guess we should focus on other behaviours - temptation, jealousy, lust, greed, etc. 

I think "forgetting" is a good example - it is generally considered to be undesirable, but I'm pretty sure it is an essential part of intelligence and the learning process. 

Maybe this can help:

You can get the similar or apropriate answer by searching the keyword in the GOOGLE SCHOLAR page. Usually you will get the first paper similar to your keyword.

From my experience, this way will help you a lot. If you still have a problem, do not hasitate to let me know.

Kind regards, Dr ZOL BAHRI - Universiti Malaysia Perlis, MALAYSIA

BNT-BKT-BT. Available from: https://www.researchgate.net/post/BNT-BKT-BT#view=54f68badd3df3ee77c8b4638 [accessed May 7, 2015].

Hi

You can think a module as a complete actuation, sensing and control.

If you are considering an industrial manipulator, then you need,

a model for each category of rotational joints (waist, elbow and shoulder joints)

a model for each category of prismatic joints (orthogonal and coo-linear)

an End effector module (male-female connection)

This mean you need 5 modules for the joints and one for the end effector.

Dear Saqib

Just to simplify some of explanations above, you can simply smooth your data by way of averaging the accelerometer output samples. That is you can add two, three or more accelerometer output samples at a time and divide by number of samples in the block to get the average each set of sample blocks. Each block or bundle average is then your new (smoothed) output sample. 

This is known as smoothing or averaging and is equivalent to low pass filtering in the frequency domain. The more samples you use to compute a new sample average, the lower the cut-off frequency of your low-pass filter, and the smoother your accelerometer data will be. 

Dont use too many samples to compute the average then, you may lose abrubt changes in your measurement that may be important. To put it in another way, too many samples in each smoothing block may result in the loss of important high frequency responses in your measurement. 

If you need formulas, see this http://www.analog.com/media/en/technical-documentation/dsp-book/dsp_book_Ch15.pdf

Regard

Gerro

Read this

alter the pwm to the dc motor for changing speeds of motor

You can use humanoid robots as a tool that can intact with children easily, it is like a human being and also can talk and also listen if it is programmed well. Now days a wide researches have been devoted to the use of this kind of robots to health care. One interesting topic is using humanoid ones in helping autistic kids. Some commercial robots have been used recently that look interesting for kids. Some scientific surveys have been done to show the research improvement and also achievements.

Hi, for a point to start. you can go through the RoboCup Humanoid League Official webpage. You can see there variety of robots which are working now well. So you can start from simple design and then go to develop them.

Hi Jawad

I am not familiar with V&V for robotics. However I would suggest the text book:

'Verification and Validation in Scientific Computing' by W.L. Oberkampf & C.J Roy

Hope you find this useful.

There are a couple of free software suites out there, in case you don't want to go into the guts of it yourself:

In case I remember another one, I'll update that post, but it should give you already a good idea.

Dear Farzin Piltan,

It has been a while since I worked with non-linear manipulators and backstepping, but in my experience it is very difficult to deal with time-delay in feedback in practice, and makes it very difficult to analyze stability for non-linear systems (I can't remember ever having seen any standard/general approach for analyzing the effect of time-delay in non-linear systems).

With regards to backstepping, I guess the more practical option is to rely on the observer you usually have to design for the backstepping control law, and the feedback of these states together with the feed-forward of the reference through the chosen dynamical model. In my experience then, the more accurate the model is, the better results you get. The simple approach is to reduce/change the gains by trial and error (simulation) in order to find an implementation that is stable (although stability can still not be guaranteed), but of course this increases the bound for the error. I guess this is the trade-off that has to be made. The robustness of backstepping (the varieties that I have seen) really do not extend to the effects of time-delay (and also, notably, does not handle sensor/actuator saturations, which can be a problem since backstepping often generates huge control signals, unless the growth rate of the control law terms is limited...)

I guess a different approach of using an Hinfinity control law (which includes an Hinfinity obeserver/generalized Kalman filter) and feed-forward linearization could be used. This might be a simpler framework to work with since it is possible to include the effects of time-delay in the linear model. Time-delay here also tends to restrict the control law gains, and depending on accuracy of the linearization, the linear model uncertainty can be large, and the two problems combined might make the performance poor, but I guess it is hard to say in general.

Changing the sensors (reducing the actual time-delay), if that is possible, does sounds like the very best option to me... Optical encoders should be really fast if used correctly. For rapid-prototyping and hardware-in-the-loop simulations, I would in general suggest using a regular PC running the MATLAB xPC real-time operating system with a National Instruments DAQ card installed (http://sine.ni.com/ds/app/doc/p/id/ds-15/), since it is usually a fast and relatively inexpensive solution that works without much modification, and since many of the DAQ cards include fast encoder inputs.

That is what I think, anyway :)

I knew the work of the KTH people with Furhat... similar to yours. Nice piece of work. Lamp avatars are a bit different from robots: the eyes are active. Sounds strange that nobody did what you did with real robots... we keep in touch.

That's what I thought too, looking at the formula. However, @Philip's answer seems to have cleared it up for me. Thank you.

Check for manufacturers catalogue and select motor which can have more than 10 kg of holding force. It also depends on what kind of accelerations will you allow. You need to give more information about it.

Autist patients and children in particular have problems related to social interaction and communication reciprocity. It was studied that for them it is much easier to interact with computer and machines, that have a predictable behaviour. Patients suffering from autism prefer repeatable actions and a much more reproducible interactions that are not always possible interacting with humans. Humanoid robots, having a human shape but being at the same time machine, can therefore facilitate the development of social reciprocity skills and interaction abilities, being at the same time a teaching and a gaming tool.

Hi Ranjan,

Here is a great step by step matlab tutorial for particle filters. It is a youtube video series of 3 parts with matlab code.

The code links are in the description of the videos.

http://www.youtube.com/watch?v=O-lAJVra1PU

Regards,

Mohamed

There are plenty of filters to deal with your problem.

The main choices are the Kalman filters (Extended Kalman filter, Unscented Kalman filter, etc) and the particle filters.

The first is useful when you have an initial position (or an idea of where it could be) and lineal (or near lineal) measurements.

The second is able to handle not knowing the initial position and having non linear measurements (map matching, etc.), but requires a significant amount of computing compared to the Kalman filters.

In all cases you will need to model the robot states (position, velocity, orientation, etc), estimate their propagation (probably using the odometry) and update the estimations with any measurement you receive.

I recommend the book "Probabilistic Robotics", from Sebastian Thrun, Wolfram Burgard, Dieter Fox on the subject.

If someone forces the monkey to control 100 robots to get one banana, I think the poor little monkey can do this.

First you need a dynamic model of your biped robot.

Then, you can use center of gravity or zero moment point techniques to begin with.

I agree with Muhammad's suggestion: KISMET. In addition you should check the results from WE-4RII robot (head and torso expression). However the main problems are the technological constraints: only 3 motors! But still the main question is "what is possible to express with only 1 DOF?" I suggest first to read the results of Expression Measurement FACS (FACS=Facial Action Coding System) from Ekman y Friesen (1978). Then, you will decide what to express.

If you want just a few commands and a not too complex system you can use LPC coefficients for speech processing. There are a lot of examples in matlab.

have a look at http://neuron.tuke.sk/~maria.vircik/danceproject/ - her mail is maria.vircik@gmail.com ... she was using EA for dance design ...