Page 1 of 2 12 LastLast
Results 1 to 10 of 11

Thread: University humanoid robot project, need controller suggestions

  1. University humanoid robot project, need controller suggestions

    I'm working in a multidisciplinary senior design group at school and we've been tasked with building a 3ft humanoid robot with various capabilities. We'll need to do object detection and avoidance, simple walking, balancing, and fall recovery to name a few. We're looking at something in the ballpark of 24 servo's total.

    I need a board that has good libraries, preferably with support for interfacing with gyro/accelerometer sensors and pwm controllers. I like the ARC-32 for servo control. Aside from that I've looked at pandoboard, roboard rb110, and a few others but nothing seems to hit the sweet spot.

    So to list that:
    - Interfacing with the ARC-32
    - Built in gyro/accel or ability to interface with separate boards that perform those functions
    - SD card storage
    - Sufficient processing power, I'm not sure what that might be though
    - Friendly documentation/libraries, possibly open source
    - Preferably wifi capable

    Any pointers would be much appreciated.
    Last edited by Jayws; 01-10-2012 at 12:03 PM.

  2. #2

    Re: University humanoid robot project, need controller suggestions

    I think you are not addressing the major design issues in the right order.

    Your target functional spec. would require really demand an embedded Linux processor, which could be be Roboard or one of many others. For now just consider that as payload. You can use ROS, DARWin, or whatever on that.

    The choice of ARC-32 is totally premature, since you dont' know what your servos are.

    A 3 ft humanoid is mechanically state of the art and probably ten times more expensive than a 1ft humanoid, are your prepared for that ($5,000 +). If so, then sort out your mechanical design, then the required servos, then the servo controller.

    Maybe whoever tasked you with this, can post some examples of their own work as inspiration to us all

  3. Re: University humanoid robot project, need controller suggestions

    Believe me, I understand the magnitude of the project. Unfortunately I don't have a background specifically in robotics so I'm not familiar with elements like the processing demands. I did some more looking around and decided the Roboard RB-100 would be a good platform to work with for a number of reasons. I also wouldn't need a controller just yet for the servo's because it has 24 PWM channels. Do I need to use compatible servo's that they recommend or can I use any servo that accepts a PWM signal? The mechanical engineers in my group mentioned wanting to use EX series servo's?

    Our professor is determined to make something that is at least 2ft tall. We recognize that this will require expensive high torque servo's. Whether he is willing to fund this or not remains to be seen as it is significantly more expensive than I think he expected when he gave us a budget. It is possible though.

    Unfortunately I don't have any examples from this teacher to post. I get the impression that this is an aggressive assignment. I've seen a hexapod and robonova in our robotics club but that's about it. The goal of the project though is to see if we can accomplish these goals and if not, leave something that a future team can pickup and continue to develop and improve.

  4. #4
    Join Date
    Jun 2011
    Location
    USA
    Posts
    547
    Images
    107
    Rep Power
    50

    Re: University humanoid robot project, need controller suggestions

    One thought that occurs to me is that you'll have an easier time if you do a biped with no arms rather than a true humanoid. This saves some weight that can be used for the sensor/battery/mpu payload, as well as saving money.

  5. Re: University humanoid robot project, need controller suggestions

    We may end up doing that to save cost, not sure yet. The Roboard RB-110 looks like the best choice to use with dynamixel ex106's as it has the high speed serial ports. It's my understanding that you can just daisy chain a single serial line to all of the servo's and send commands to them with <= 18ms response time.

  6. #6
    Join Date
    Jun 2011
    Location
    USA
    Posts
    547
    Images
    107
    Rep Power
    50

    Re: University humanoid robot project, need controller suggestions

    The high speed shouldn't matter - I've not yet heard of a board that can't do the UART stuff at 1 mbaud. (I think there are some software tricks that make this work, but that's something that whatever code library you use takes care of for you.) (also, UART is basically what we call serial ports on micro controllers *shrug*)

    You are correct on the daisy-chaining.

  7. #7
    Join Date
    Sep 2010
    Location
    ಠ_ಠ
    Posts
    2,317
    Images
    27
    Rep Power
    285

    Re: University humanoid robot project, need controller suggestions

    Summarizing iBot above (and every reputable engineering design course):
    "You have to know what you want to do to figure out how to do it".
    Until you define your goal and attributes, don't settle on any hardware. While you are compiling your list of potential/interesting/marginally-related hardware, be very sure to well-research them for any potential conflicts with the final project goal.

    The mechanical design will dictate servos/actuators which will in turn dictate communication/control methods for the servos (PWM is used mostly in cheap-ish hobby servos, EX-106 uses RS-485 half-duplex UART, and MX-106 will use TTL half-duplex UART).
    The overall project goal/purpose of the bot will dictate what sensors and processing architecture you use (do you want to use Windows, Linux?; do you want to run ROS, OpenCV, PCL?, etc.).
    If the two do not meet/overlap, you will need an intermediate board like the arbotix or ARC-32 to control the servos and read sensor data over USB.

    If you know that you want to use ROS, then know that you will need an x86-compatible processor (ARM is not terribly well supported) (preferably with all modern extensions like SSE/SSE2/SSE3 for OpenCV/OpenNI/Kinect/etc.; the Vortex86 on the roboard implements only i586 without SSE) (easiest OS for ROS is running Ubuntu Linux) and at least 8GB of hard disk memory.



    [cue pedantry]
    UART is a very common serial data transmission protocol that uses pre-agreed upon clock frequencies and transmission formats (8N1 being most common - 8 bits data, one stop bit, no parity bit) (line defaults high. pull line low for one clock cycle to indicate start bit, toggle line for 5/7/8/9/10/etc clock cycles to transmit data bits, toggle line for parity bit (if used), pull line high for 1 or 2 cycles for stop bit(s) before starting next transmission). TTL (usually 0-5V), RS-232 (+/-5V or greater), and RS-485 (-7 to +12) are electrical specifications (RS-232 standard includes several additional line signals to assist in transmission and other triggers).
    Please pardon the pedantry... and the profanity... and the convoluted speech pattern...
    "You have failed me, Brain!"
    [git][mech][hack]
    gives free advice only on public threads

  8. #8
    Join Date
    Jun 2011
    Location
    USA
    Posts
    547
    Images
    107
    Rep Power
    50

    Re: University humanoid robot project, need controller suggestions

    Quote Originally Posted by tician View Post
    [cue pedantry]
    UART is a very common serial data transmission protocol that uses pre-agreed upon clock frequencies and transmission formats (8N1 being most common - 8 bits data, one stop bit, no parity bit) (line defaults high. pull line low for one clock cycle to indicate start bit, toggle line for 5/7/8/9/10/etc clock cycles to transmit data bits, toggle line for parity bit (if used), pull line high for 1 or 2 cycles for stop bit(s) before starting next transmission). TTL (usually 0-5V), RS-232 (+/-5V or greater), and RS-485 (-7 to +12) are electrical specifications (RS-232 standard includes several additional line signals to assist in transmission and other triggers).
    I'll go right ahead and admit I was attempting to bait someone into giving a good explanation. Thanks

  9. #9
    Join Date
    Sep 2010
    Location
    ಠ_ಠ
    Posts
    2,317
    Images
    27
    Rep Power
    285

    Re: University humanoid robot project, need controller suggestions

    Quote Originally Posted by Gertlex View Post
    I'll go right ahead and admit I was attempting to bait someone into giving a good explanation. Thanks
    Yeah, it is a source of great frustration (and greater confusion for the uninitiated) when someone use RS-232 as a generic term for serial communications (an vice-versa). How I loathe those 9-pin connectors.
    Please pardon the pedantry... and the profanity... and the convoluted speech pattern...
    "You have failed me, Brain!"
    [git][mech][hack]
    gives free advice only on public threads

  10. #10
    Join Date
    Feb 2008
    Location
    California
    Posts
    421
    Images
    33
    Rep Power
    57

    Re: University humanoid robot project, need controller suggestions

    A humanoid standing at 3ft, I think the ARC-32 choice is really a bad one. The heftiest servo I've seen that uses the standard hobby-type/size is in the mid 400 oz/in.

    What you are describing, you would need something like the EX (or the soon to be released MX) 106 servos. Even then, I'm not sure how successful you would be, since ultimately, anything Hitec or Robotis is, in the end, hobby grade stuff that can remain only to a certain scale.

    There's a point where servos become a diminishing return, that is the torque to weight ratio. Each servo has its own torque to weight ratio, but the real equation is the weight of the entire bot (standing at 3ft) versus the holding/stall torque of a single servo, since I imagine that unless you double up on the servos at each joint, for some conditions/poses, that servo will be carrying the entire weight of the bot.

    If you are looking to build a 3-foot robot, I highly suggest start looking at some of the mechanical and the electrical systems of existing university/industrial grade humanoids like the Honda Asimo, HRP (sp?), Aldebaran, and etc.

    If you don't mind me asking, what is the actual objective of this University research? Is this simply building a bot to build a bot, or do you actually have a topic or topics that you are trying to research/solve?

    Like others have mentioned, define your objectives, then derive the requirements (mechanical, electrical, software), then it may be easier to see what you will need.

    If you want my opinion, if you just want a PWM based servo controller, I believe the ARC-32 is a bad choice. SSC-32s are specifically geared towards that and it does nothing but that, ARC-32's does do servo controls, but it requires a bit of coding. The SSC-32 has a nice serial interface and many of the low-level stuff like timing is taken care of by the unit itself.

    I love the ARC-32 for its size and the ease of use of the SDK that Basic Micro provides, but it does have its limitations. A humanoid robot is one of the toughest, you not only have to do the "standard" bot programming, but you will also need quite a bit of mechanical finesse to get it right.

    Not trying to put you down, just need to define your goals, and if I am to assume that this is an ENGINEERING project, define some realistic goals so you don't end up wasting a bunch of time and money...
    ---
    Sleep? You don't need sleep...

    ---

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

Similar Threads

  1. Question(s) University Senior Project-Xbee?
    By bcompto5 in forum Robotics General Discussion
    Replies: 0
    Last Post: 09-26-2010, 12:11 AM
  2. Question(s) Motor Controller suggestions
    By Mike in forum Arbotix, Microcontrollers, Arduino
    Replies: 0
    Last Post: 04-28-2010, 10:43 AM
  3. Interesting new humanoid robot, what do you think of it
    By openmindedjjj in forum Robotics General Discussion
    Replies: 78
    Last Post: 09-08-2008, 02:55 PM
  4. New Project parts suggestions needed.
    By misiu in forum Robotics General Discussion
    Replies: 11
    Last Post: 02-01-2008, 12:58 PM
  5. humanoid pico-itx robot
    By asbrandsson in forum Project Showcase
    Replies: 35
    Last Post: 02-01-2008, 06:26 AM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •