The Research And Implementation Of Target Recognition Algorithm And Action Planning In Humanoid Robot

Humanoid robot is the future direction of robotics, which has the ability to imitate humans in the structure and control level. Humanoid robot is a concentrated expression of theory and technology in intelligent robots. However, the development in target recognition and walking stability is still unsatisfactory. Under the analyzing of existing target recognition algorithm and walking gait model, combined with the advantages in hardware processing, the FPGA-based target recognition algorithm and robot walking stable gait model in target finding are proposed.Firstly, based on the characteristics of small humanoid robots, the deficiencies of common color compression method are analyzed and improved, the FPGA-based target recognition algorithm is designed. The algorithm uses look-up table based color compression methods and detects the target with statistical method, which innovatively adds the asynchronous feature to the module, thus implements the parallel computing of RGB receiving module, two-dimensional filtering and data compression module and target color comparison module.Secondly, based on the recognition of target, according to the behavioral characteristics of humanoid robots, the method of ZMP stability algorithm is improved, walking stable gait model in target finding is designed .The model uses FPGA-based target recognition algorithms, utilizes the kinetics principle of gait and implements the walking stability in robot walking.Finally, the Verilog hardware description language is used to realize the proposed recognition algorithm, and the ModelSim-Altera simulation software is used to commit the timing analysis of target recognition results; The gait model of robot walking is simulated under Matlab 7.0 platform, which verifies the stability of gait planning.