Tuesday, March 17, 2009

Last Day In Korea

Well Happy St. Patrick day for starters and Departure date from Korea! I leaving Korea around 11am to be back in the States 14 hours later...I'm going to miss Korea!!! :(

Friday, March 13, 2009

The Manufacturing - Robogames 2009

"Double Knee" Final Design Robogames 2009

This is the Final CAD lay out of our Robot for the 2009 Robogames Competition. I decided to go with the Double Knee configuration , for the proposed running attribute. The Chassis Specification are 1.7 lbs machined parts, 51cm tall, 24 DOF. Incorporated in the feet are also custom XYZ Force Torque sensor for foot landing data collection. The colors are the proposed Anodizing scheme, but that will come later. I know what you want to see the real thing right? Wait for the next post...the Manufacturing!

Thursday, March 12, 2009

Prototype "Bent Knee"

This is the "Bent Knee" approach I prototyped. Notice the distinct feature of the the knee motor is set a 45 degree angle normal to the ground. This I believed allows the robot to lift its leg to take a step with out it needing to swing the lower half of the knee out as far in order to progress forward. Also the ankle pitch does not need to rotate to maintain keeping the foot parallel to the ground. The design uses what I defined as stacked axes , which is the pitch and roll are offset from each other in height. Thou being a clever set up of the motors, that was the down fall why we did not go with this design. As a team we decided we wanted all pitch and roll to be fall on the same plane for the hip and ankle motor orientations. Below is closer look into what I described as the Gate motion of this prototype. I guess it is shelved for now but I may want to revamp it for a future candidate.

Saturday, February 28, 2009

Last few Weekends in Korea - Pool Hall

This weekend I decided to go out on Friday with one of my good friends from Uzbekistan. We went to a local pool hall to shoot around. I had a good night I won 4 of the 6 games we played. Here is an action shot from last night!

The Dynamixel

To build a mini robot you need to have small motors to drive the articulation of each joint. In Hubo's case It uses a complex set of multiple motors and drive belts in combination with Harmonic Drive systems to produce movements and have holding torque to stand and move. This method work wonders for Hubo but is not a good system for a small system for reasons that it is bulky and very expensive. On top of the mechanical complexity it needs to have custom motor controllers to drive the motor train as well. Small Humanoids can use a off the self package that includes a motor, gear set, encoder, and controller. These servos are far more cost effective, light weight, and have far less overhead to program the included controller. The servo package we chose for this project is the Robotis Dynamixel. Robotis is actually a Korean based Company that sells robot kits such as the Bioliod Humanoid. They sell many other accessories and servos for people interested in building custom Humanoids like ourselves. We deiced to go with the RX-28 that supplies 28N.m of torque for leg servos and RX-10 (10N.m) for the upper body.

Robot Inspiration

I spent many hours searching internet based videos to see what types of designs existed for small Humanoids. Granted I know the design of Hubo inside and out but yet this robot could not be as complex. So I needed to "switch gears" to start thinking in the perspective of small robots. I found two robots that I liked that were featured in the ROBO- ONE, a Japanese competition. The first robot that gained my interest was the Kondo KHR-2HV.
I liked the legs where the knee joint is set on an angle to the perpendicular part of lower leg and foot. This allowed the robot to have a very fluid motion during walking. I also felt it did not have to Throw its leading and trailing leg as far to achieve locomotion.

The second design that also took my interest was a small robot also from ROBO - ONE. This robot is called OmniZero.
The video shows the unique construction of the robot, notice that this robot has two motors at the knee joint. The double knee is much different than most Humanoid robots competing. Most robot have a single motor actuating at the knee, but in using two motors in the knee allows each motor to rotate half the distance. This allows of for faster actuation at the knee and I believe that this is a potential design for a more capable running robot.