|
| Home / Projects / Land |
 | Bi-Ped Project
Jaap Havinga. 20 Nov 2000
|
Project Outline I started this project to find out how much it takes before I have constructed a robot that walks like a chicken. The aim
is to build a compact robot that is capable to manoeuvre in common indoor conditions, without the requirement of
operating on a smooth floor. I will describe the development in terms of increments. Several increment descriptions
include video-clips of the walking Bi-Ped. |
Latest Journal Entry| Increment #4 30 Nov 2000 | | Enforced hardware, Servo controller doubled in precision, Force feedback.
Hardware
On the previous version, if the robot stands on one feet, the standing leg was bending over more than 2 cm inwards due to the weight applied on it. This was overloading the servo-motors and made the whole construction unstable. Therefore, the first thing to do in this increment was to improve the lateral stiffness of the legs by adding extra ball-bearings at each hinge, see picture. Now the bending is reduced to about 0.5 cm and seems to be acceptable.
Servo precision
The next thing I improved was the precision of the servo-guidance. It appeared that a source of the somewhat shaky movement of the legs was due to the stepwise movement of the servo's: the step-size was about 1 degree. The servo controller was reprogrammed and the current precision is 0.5 degrees. This greatly reduces the trembling.
Force measurement
The most important step was to enable measurement of the applied forces on the legs. The measurements are feedback into the system in order to let the biped walk on non-flat floors. The simplest way to obtain a static force-measurement appeared to a tap on the controlling signals in the servo itself. If a servo has to maintain a certain position while a force is applied on the servo-axis, the servo-controller will apply some current on the dc-servo-motor, such that the dc-motor produces a equal amount of force in opposite direction. So the pulses applied on the motor are proportional to the applied (static) force on the servo. The servo controller generates mutually excluding logic pulses for driving the motor leftwards and rightwards, see figure 2 and 3. (These pulses are input to a H-bridge that is driving the motor itself).
I installed another PIC controller to measure the amount of servo-pulses for 6 servo's. The measurements are sent every 60ms to the CPU by serial interface.
Force feedback
The last thing to do was to integrate the force measurements into the biped controlling system. For this, every servo is controlled by a PID-controlling object, and has different parameter settings for the different stages of the leg: standing or moving forward. The control of a standing leg is such that the leg remains in upright position. The control of the moving leg concentrates on the alignment of the feet on the floor at the time the feet will settle on the floor.
|
Other Journal Entries
| |
|
|
