Bipedal walking robot(2L1v6)

I'm wishing to build bipedal walking robot since I saw bipedal walking robots in early '80. And I started to build one when I saw P2 (HONDA's humanoid robot) on TV in '98. So, I thought "It's time to start my project". This robot is under construction.

The meaning of number after the name of the robot is the version number. I change the number after big modification.(e.g. change of D.O.F) The version "v6" is the last.

Software environment

There are two kind of software to control this robot. One for PC and the other for micro controller that is included to the robot. I use C language for programming. And the programming environment that I'm using is Microsoft Visual C++ Ver. 6. Of course the native compiler for the chip compiles the program for the micro controller after checking of syntax by VC++ Ver.6.

The software for the PC plans the path for the legs and calculates the data set of the angle for each of the joints. And the software for the micro controller generates the signals to drive RC servo motor and collect data from the sensors and send them back to PC.

2L1v6

	Walk forward:2002/07/21 The robot walks forward and turns. The robot is controlled like RC model car in this movie. The micro controller and battery is included into the robot. No connection with PC. (1371 Kbyte/mpeg)
	Walk aside:2002/06/19 The robot walks right and left. (344 Kbyte/mpeg)

2L1v5

	forward 4:2001/07/08 The robot steps ahead after the motion of swinging is settled. The robot detects the motion of swinging with rate gyro that is equipped in the body. (232 Kbyte/mpeg)
	Step aside 2: 2001/04/07 The robot steps aside. This is the first shot that I have changed the algorithm to calculate the orbit for the feet. I give one orbit to hip and offset to each foot after this shot. (I give the orbit to each foot before. ) (232 Kbyte/mpeg)

2L1v4

Knee bending test: 2000/11/12 Knee bending test is my basic testing method. I'm adjusting geometrical parameter and see the balance of the robot without stabilize program that use gyro and acceleration sensor. (120 Kbyte/mpeg)

2L1v3

	forward 3: 2000/08/08 Static walking with balancing program. But it seems that the program is not working. I think that the response of the servo motor around the foot is not enough. (232 Kbyte/mpeg)
	step aside 1:2000/08/24 Step aside with static walk. (232 Kbyte/mpeg)

2L1v2

	the first step:2000/03/15 The robot get speed around roll axis by bending knee slightly. (120 Kbyte/mpeg)
	forward 1: 2000/03/12 The robot steps 6 steps. It about to fall over. The function to control the motion around roll axis does not work well. So the direction to the desk is fixed. If the direction is changed, the robot cannot walk form the inclination that I cannot see. (344 Kbyte/mpeg)

Photos

	2001.02.17 2L1 Version 5 The DOF is increased from 10 to 12. And a battery for the servo motors is included in the head(torso?).
	2000.11.04 2L1 Version 4 I have changed the mechanism of the ankle. The size of the servos are changed form mini type to normal type.
	2000.04.15 2L1 Version 3 The DOF is 10 at this point. Some kind of parallel link mechanism is applied to the joint of the ankle. The shape of the contact to the ground is changed from line to plane. The robot could get to walk statically with this modification. But the servos are broken from over load. It seems that the torque was not enough to support the weight of the robot.
	1999.11.7 2L1 Version 2 The DOF is 8 at this point. I put a rubber band on each leg to stabilize the vibration of the legs.
	1999.07.20 2L1 Version 2 The system of this robot. From left to right, body of the robot, signal distributor, servo controller and parallel port adapter.
	1999.05.23 2L1 Version 2 Inverse kinematics program is tested with hanging the robot. I always hanging my biped robot for servicing too.
	1999.04.24 2L1 Version 1 The first stage of dynamical reaction test was done with two DOF. The legs in front of the robot are the legs for version 2.

Detail photo of the structure

This robot is mainly made from plastic plate.(t=1.0mm) And the joints are reinforced with aluminum plate. (t=0.5mm) The base for every servomotors are made from aluminum angle.
(Click to see larger image)

	Body This photo shows the inside of the body. Battery, acceleration sensor, angular velocity sensor (rate gyro) are included.
	Hip The degree of the freedom of hip is three. The layout of each joint is orthogonal, but do not intersect at one point. This layout requires more calculation power compare to the joint which all three degree of freedom intersect at one point. But I don't think it is too heavy for recent PC.
	Knee The knee can be bend only about 45 degree. I think this is not enough but the actuator has not enough power. So, I think that to increase the torque by the combination of the links takes priority over the range of this joint.
	Ankle(roll axis) The degree of the freedom of ankle is two. The layout of each joint is orthogonal, and intersect at one point.
	Ankle(pitch axis) The base piece of the servomotor is made from aluminum angle bar. And glue onto the leg's structure.
	Shin cover The shin cover is attached to the shin structure with eight screw.

Mechanical Drawings

Click A-G to see the drawing. A:Torso B:Hip1 C:Hip2 D:Thigh E:Shin F&G:Ankle & Foot