Control System Bottom Design And Gait Planning Of Humanoid Robot

Humanoid robots'soccer game is a confrontational game of robots which has developed rapidly in recent time. Building the platform of humanoid soccer robots and participating in this activity can help to promote the research in the application of embedded theories and artificial intelligence.Based on the original humanoid robot AFU in the laboratory, this topic designed an improved humanoid soccer robot AFU2008, which was aimed to get higher score in the match. The thesis mainly covers the following three points:1. Firstly this thesis analyzed the design provisions of humanoid soccer robots in RoboCup activities. Combined with the experience of the original robot AFU, it designed a brand new mechanical structure and chose more advanced servo motors. After this, detailed analysis of the control system's demands was carried out and this thesis selected master-slave architecture as the control system: the motherboard used Samsung S3C2440A as the processor, which completed the upper tasks such as real-time visual information processing, wireless network communication, decision-making and task planning. The motion control board chose DSP TMS320F2812 as the processor, which accomplished bottom tasks including sensor information collection and transformation, the control of the servo motor. RS232 interface was adopted to carry out the communication between the motherboard and motion control board. Then the specific circuit of the control system was designed and its principle was analyzed in detail.2. At first, the advantages and disadvantages of several major operating systems were analyzed and compared, finally embedded Linux with more advantages was selected as operating system of the robot AFU2008, and the kernel version is chosen to be 2.6.13. Then the work principles of BootLoader were introduced, and VIVI was used to be the load program of the control system and its modification and transplantation to the board were completed; after that, cross-compilation environment was built, and kernel and root filesystem were modified and transplanted to the board. At last, the Linux system on the board was started, and the main function modules were tested due to their necessity.3. This thesis established the kinematics model of humanoid soccer robot; after this, according to the ZMP (Zero Moment Point) stability criterion, gait planning of the robot was carried out by using the method of referenced cosine trajectory; and the values of joint angles which have been solved out through inverse kinematics were then sent to control the mini servo motors of the humanoid robot. Meanwhile the real-time actual ZMP was measured and calculated through the using of pressure sensors on the feet, and then compensation method was applied to carry out on-line correction according to the error between actual ZMP and desired ZMP. At last, effectiveness of this method was proved through simulations and experiments.