報告題目: Experimental estimation of tactile reaction delay during stick balancing using cepstral analysis
報告人：Dalma Nagy （Department of Applied Mechanics of Budapest University of Technology and Economics金世豪娱乐_首页，Ph D Student)
The neurological mechanism during human balancing tasks is often modelled as a computer controlled delayed feedback process. Using the sensory organs humans are able to detect changes in the state of the balanced object. These information are then transmitted to the brain, which orders the muscles to act all in a synchronized process that maintains balance. Stick balancing on the fingertip is a simple balancing task, however, the investigation of this task can help to understand the complex control mechanism of the human brain. In this work, the human controller during stick balancing was analysed with experiments and the results are compared with simulations. The mechanical model of stick balancing was developed and it was assumed during the research, that humans can detect the position, velocity and the acceleration of the stick. Therefore, a PDA type controller was used for modelling the controller employed by the human brain, where PDA stands for position, velocity and acceleration feedback. The controller operates with a delay, since it takes finite time for the neurons to transmit information to and from the brain. The effect of sensory threshold was also taken into account during the modelling process, thus, a switching-type controller is applied. Measurements were carried out to obtain estimates for sensory thresholds. The measurements consisted of two separate experiments for the determination of the position and velocity threshold. By applying a time delayed PDA controller, the mathematical model describing stick balancing results in a neutral delay differential equation (NDDE). For systems governed by NDDEs, cepstral analysis can be used to identify the time delay, and to gain information about the neutral behaviour of the model. The usability of cepstral analysis was tested on simulation results first: cepstral analysis of signals obtained by time-domain simulations of the model of stick balancing was performed, and the effect of changing the value of the acceleration feedback gain on the cepstrum was analysed. Measurements were carried out in order to verify the proposed model of the human controller by checking, whether hints of acceleration feedback can be detected in the cepstrum. Cepstral analysis of the measured time signals was carried out and an estimation for the time-delay of the acceleration feedback was determined.
Dalma Nagy is a Ph D student of the Department of Applied Mechanics of Budapest University of Technology and Economics. Her research topic is Modeling and experimental investigation of human motion control in balancing tasks. The topic involves mathematical modelling of human balancing tasks and carrying out measurements in order to better understand the neural mechanisms behind balancing.