| Accurate recognition of unknown terrain by a robot plays a vital role in the robot’s ability to complete a given task.Currently,the environmental information perception methods commonly used in the field of terrain recognition research suffer from severe environmental influences,low accuracy,time consuming,high cost and limited applications,while the use of haptic system to perceive terrain can solve this problem to a certain extent,so it is important to study the use of haptic system to perceive terrain for terrain recognition.However,as the complexity of robot application scenarios increases,there are still many technical difficulties in using haptic perception systems for terrain recognition,among which specific techniques such as using haptic perception to improve terrain recognition accuracy and shorten computation time are in need of breakthrough.Therefore,this paper presents a new terrain recognition model utilizing the probabilistic neural network algorithm in a haptic force feedback system.A pressure sensor is mounted on the foot of a hexapod robot as the front-end sensing unit of the haptic system to collect motion pressure feedback information,and these data are processed to extract DEM feature values and finally terrain recognition using a probabilistic neural network algorithm.This paper presents experimental verification of the algorithm’s accuracy and speed.The average terrain recognition accuracy of the model is 96.7%,and the calculation time is 0.03～0.09 s.In addition,to visualise the performance analysis of the terrain recognition algorithm,an upper computer software system for terrain recognition is designed based on Matlab/GUI to represent the data and recognition results in the form of an interface.The research on terrain recognition algorithm of hexapod robot based on tactile force feedback system is mainly divided into the following aspects:1.In this paper,the environment perception method based on tactile system is proposed and compared with the current common environment perception method.Select slope,slope change rate and aspect as characteristic values.In the process of terrain recognition,terrain feature values are extracted for terrain recognition.2.This paper studies the design of haptic feedback control system for hexapod robot based on ARM chip STM32F103VET6.The main components comprise the central control level,steering gear drive layer,power supply layer,and environmental information sensing layer.This paper provides a detailed account of each module’s device selection,circuit design,and function implementation.This paper introduces several classical gait of hexapod robot and its related technical parameters,and carries out gait planning design and gait phase research.In addition,this paper designs a tactile force feedback system,which is composed of pressure sensor FSR400 and main control board.Communication is realized through serial port.It can provide tactile force feedback information when the robot crosses different terrain,thus realizing terrain recognition.3.This paper studies the terrain recognition algorithm of hexapod robot,and proposes a terrain recognition algorithm based on probabilistic neural network.On the basis of selecting terrain slope,terrain slope change rate and slope direction as terrain feature values,three classification algorithms for hexapod robot terrain recognition,namely probabilistic neural network(PNN),error reverse neural network(BP)and limit learning machine(ELM),are trained respectively.By comparing the recognition accuracy of the three trained models,the superiority of probabilistic neural network is verified.4.In this paper,the terrain recognition algorithm based on the hexapod robot foot sensor is tested.Gait planning experiments were done first,and then terrain recognition experiments were done.Experiments confirm the effectiveness of the algorithm.5.To achieve the visualization of performance analysis for terrain recognition algorithms,the hexapod robot’s terrain recognition algorithm model is simulated and analyzed.The Matlab/GUI simulation platform is utilized to design the simulation interface for hexapod robot terrain recognition. |
