About
My name is Peter Neal. I live in Cornwall in the UK.
I have been building robots for many years after obtaining a degree in robotics and automated systems. I will be posting updates on my latest robot projects and you will also find information about my previous projects.
Big Face Robotics 
May I ask: are you USA-based or elsewhere?
I’m in the UK
I thought so. Nice work.
Mr. Neal, please email me at Maxwell.B.Rebo@gmail.com, in regard to your robot. Thank you.
Hi! I really liked your Humanoid robot a lot, I am a graduate student and am making a humanoid robot with speech to text & text to speech capability to make it more realistic as my final year project. I must say your work have motivated me to go even further. I had one small question….why are you using 3 servos on the shoulder? Wont 2 be enough? & which servo are you using? Whats the material of the structure? I am planning for thin acrylic….It would be very helpful….Thanks!
Hi, thanks for the kind comments. It’s nice to know its worth sharing my work if it gives other people ideas and a little inspiration! In response to your questions, yes you could use just two servos for each shoulder joint but you will lose a degree of freedom. I think using 3 servos gives a much better range of motion than using just 2. It depends what you are trying to achieve. If you are making a walking humanoid and weight is an issue, 2 servos per shoulder may suffice. The servos I’m currently using are Tower pro servos, nothing fancy. The first shoulder joint required a servo with more torque than the rest so I’m using Tower pro digital servos for that particular joint. The robot is constructed from aluminium as its light weight and strong. Hope this helps.
Thanks for the quick reply! Am actually planning on 2 servos for the shoulder due to severe financial restrictions. For the shoulder base am planning for torque around 15 kg/cm n rest around 8 kg/cm…would that be enough? I guess I’ll be going with thin acrylic or aluminium which ever is easier for me to work upon, i was also thinking about laminated depron for the body as its very easy to work on using very simple tools… My main aim is to add some artificial intelligence using chatterbox API’s & stt & tts capability and a bit of image processing like yours..:) Am currently badly struck with the mechanical part due to very low skills. I have a robotic arm which i made as a hobby project around 2 years back so might use something like it. In case if you are interested check it out http://www.youtube.com/watch?v=u6oTloTJHPA
Nice robot in the video! I think 8kg/cm and 15kg/cm servos would be more than enough but I’m not sure of your exact requirements. Most of the servos on my robot are 4kg/cm but good quality servos with 3kg/cm torque work just fine. The first shoulder joint needs a bit more, anything over 5kg/cm seems to work well. I have designed the arms on my robot to be light, if you are having a gripper on the end of your arm then your servo torque requirements could be considerably higher. The best advise I can give you regarding mechanical design, particularly material choice, is to experiment and learn the advantages and limitations of each material. Then you can make an informed decision on which is the best material for each part of your robot. Most materials, metals included, can be shaped and worked with simple hand tools. Hope this is useful advise.
Its more then useful advice…thanks alot 🙂
Yeah i have already started with experimenting with various materials. My scope is very limited in terms of mechanical capabilities (no gripper) my main aim is to include some AI & some image processing. Just a small question. Are you using any servo shield on your arduino mega or direct interfaced them?
I have designed and made a custom interface board to connect servos, potentiometers and some switches to the Arduino mega. This board is connected via wire assemblies to the Arduino as the interface board is mounted higher on the robots body. I designed it this way to avoid having to extend all of the servo cables.
Thanks for such valuable advice and help…:)
I was wondering if you could help me a bit with the mechanical part would be very helpful! Thanks! amirajdhawan@gmail.com
hi, i am a fan of your robot crum, i’ve been reading the bibliography you recommend, but i just have some questions about the occupancy grids method that i can’t answer, can yoy help me? My questions are: how do you know how many grids you have to use?, does it depends on the sensor you use? Thank you for your time.
Hi, the number of cells you use in the grid depends on several factors and ultimately determines the resolution of the map. The size of the robot doing the mapping will affect the grid resolution you choose. If your robot is 50cm wide and your grid resolution (cell size) is 1 meter for example, there is a chance the robot may map an opening that it could fit through as an obstacle. Therefore it would make sense to choose a resolution as small as possible to give a more accurate map. However, a very high resolution grid/map will require more processing power and memory. Real time mapping needs the grid to be updated as quickly as possible and too many cells in the grid may take too long to process. The size of the area to be mapped will also have a impact. From a memory point of view, it may not be feasible to map a square kilometre with a cell size of 1cm x 1cm. 10000000000 cells is quite a lot of cells.
It may be these constraints that ultimately determine what resolution is feasible. Personally, I would choose a cell size around a quarter of the width of the robot to start with, and use this to work out the width and the length of the grid (in number of cells) required to cover the area you are mapping. You can always decrease the cell size and thus increase the number of cells in the grid at a later date if required. Hope this is helpful.
Yes it is very very helpful, thank you very much for answering and for sharing your knowledge. Greetings from Mexico.