Moving Target Tracking And Optimal Trajectory Estimation Of Mobile Robot Based On Auditory Perception

With the rapid development of computer and signal processing technology, targettracking has been widely applied in fields such as military air defense, traffic control andmachine intelligence etc. Target tracking is using filtering algorithm to estimate the motionstate of target, and the core part is filtering algorithm. The traditional kalman filter andextended kalman filtering algorithm will lead to declined accuracy and even filteringdivergence in the research of nonlinear and non-gaussian dynamic system. Particle filteralgorithm is applicable to any nonlinear, non-gaussian system in theory, but there is seriousparticle degradation problem. So, based on research of particle filter, one improved particlefilter algorithm is put forward to realize the dynamic tracing of sound source. In addition, inview of the optimum trajectory of robot, a method based on particle swarm optimizationalgorithm is proposed.First of all, theoretical basis of auditory perception is introduced, focused on soundsource localization method and formula deduction for geometric positioning of the fourmicrophone array; And then principle of target tracking is presented, three kinds ofnonlinear filtering algorithm are studied in detail, tracking simulation experiment is carriedout and advantages and disadvantages of each algorithm are analyzed.Secondly, to overcome the disadvantage in particle filter tracking process, oneimproved particle filter algorithm based on particle filter is put forward. The improvedmethod is to generate the importance density function using the second order centraldifference filter and also add Markov moving steps to re-sampling process. The improvedalgorithm is applied to mobile robot sound source target tracking, and the simulation resultshows that the improved method had much higher precision and could realize targettracking more accurately.Finally, in view of the fact that robot sound source localization technology can onlyrealize orientation of sound source in practical application and does not get any distanceinformation, sound source localization technology is combined with mobile robot motioncontrol, based on the active motion of robot, using particle swarm optimization algorithm toobtain the robot’s optimal detection point, so that location tracking of sound source can bemuch accurate.