| 摘要: | 研究期間：10108~10207;This is a three years project. To generate a stable basic motion trajectory for each motion to a humanoid robot is the objective of the first year. The humanoid robot can follow the trajectory to do the corresponding motion stably. These motions include bowing, kicking one foot forward or sideward, raising one hand forward or sideward, etc. The trajectories are generated by computation and simulation, then let the humanoid robot perform the motion by following the corresponding trajectory to verify the stability performance. The process of trajectory generation utilizes zero moment point (ZMP) concept and fuzzy logic to design the control of ankle angle, and then the trajectory is modified by linear interpolation and function approaching methods. Therefore the trajectory can apply to the robot such that the robot can do the motion stably and smoothly. In the second year, we will keep generating more trajectories for the more motions, such as standing up, squatting, walking forward or backward, body turning, going up or down stairs, etc. Then the all motion trajectories are collected into a data base such that the robot can do any motion in the data base stably. Furthermore, we paste 13 color marks in the human body and then recognize and record the motion trajectory when the human moving with color marks. Therefore, the human motion trajectories can be referred to generate the initial trajectories which will be modified by the technique of the first year to generate the final trajectories. With the aids of color mark recognition, the robot can imitate human complicated motions. In other words, after the robot sees a humans dance, then it can imitate the human to dance. In the third year, some pressure sensors are set under the soles of the robot’s feet to improve the motion stability control of the robot. Moreover, we will establish two man-machine interfaces, one is for users to generate motion trajectories for different humanoid robots, and the other is for users to choose several different motions and arrange them in order, then the robot can follow the arranged motions order to do the motions. |
