Software Design And Optimization Of A Perceptual State System For Quadruped Bionic Robot

Complex sources of perceptual information, high requirements in information complicity, real-time data communication and reliability are the issues always concerned, but hard to be tackled, in the development of multiple legged bionic robots. The paper provides a new design of perceptual state system for quadruped bionic robots. The system can real-timely monitor robot’s movement posture and work state of each unit to obtain their basic characteristics with the help of multi-source information fusion processing technique, which then provide real-time and key body information for planning coordination layer using for path planning and balance control.Firstly, requirement analysis of the perceptual state system for quadruped bionic robot is determined to confirm its function, target and overall program. Then the software architecture is designed to provide a sound basis for its software design.Secondly, by analyzing the characteristics of perceptual state information of quadruped bionic robot and the requirements of detection and process, data acquisition and data processing layer are established, then a double-core architecture for real-time information collection and data analysis is introduced, and the software functional modules of each layer are fully designed.Thirdly, to realize high real time of perceptual state system for quadruped bionic robots, an optimization method based on NIOS Ⅱ soft-core is presented, such as related optimization techniques based on CPU configuration and program code, as well as some optimization methods by using the custom instruction and the c-to-hardware (C2H) acceleration techniques, which are applied in attitude estimation and collision detection algorithm. The experimental results are to verify the validity of software optimization.Finally, Software performance of perceptual state system for quadruped bionic robots concerning multiple data acquisition indicators, information processing time indicator and collision detection indicator is tested to verify whether it meet the requirements of planning coordination layer.