Multi-loop Pilot Behavior Identification in Response to Simultaneous Visual and Haptic Stimuli

Abstract

The goal of this paper is to better understand how the neuromuscular system of a pilot, or more generally an operator, adapts itself to different types of haptic aids during a pitch control task. A multi-loop pilot model, capable of describing the human behaviour during a tracking task, is presented. Three different identification techniques were investigated in order to simultaneously identify neuromuscular admittance and the visual response of a human pilot. In one of them, the various frequency response functions that build up the pilot model are identified using multi-inputs linear time-invariant models in ARX form. A second method makes use of cross-spectral densities and diagram block algebra to obtain the desired frequency response estimates. The identification techniques were validated using Monte Carlo simulations of a closed-loop control task. Both techniques were compared with the results of another identification method well known in literature and based on crossspectral density estimates. All those methods were applied in an experimental setup in which pilots performed a pitch control task with different haptic aids. Two different haptic aids for tracking task are presented, a Direct Haptic Aid and an Indirect Haptic Aid. The two haptic aids were compared with a baseline condition in which no haptic force was used. The data obtained with the proposed method provide insight in how the pilot adapts his control behavior in relation to different haptic feedback schemes. From the experimental results it can be concluded that humans adapt their neuromuscular admittance in relation with different haptic aids. Furthermore, the two new identification techniques seemed to give more reliable admittance estimates.

Direct and Indirect Haptic Aiding for Curve Negotiation

“…It should be noticed that a driver with stiffness close to infinity, that is keeping the steering wheel exactly in the center, would produce the same vehicle trajectory that the DHA system produces when the pilot is not holding the steering wheel (i.e. when he has zero stiffness) [13]. The IHA controller, in its internal structure, was designed in the same way as the DHA, except for the force gain of the haptic subsystem which must have the opposite sign; this way, the force feedback commanding the steering wheel is perceived as a disturbance by the operator.…”

Identifying time-varying neuromuscular system with a recursive least-squares algorithm: a Monte-Carlo simulation study

“…The dynamics of the whole neuromuscular system are determined by the settings of these components. Humans are able to adapt the neuromuscular dynamics to the task they perform [2]- [4]. In realistic control situations, e.g., during postural control or driving tasks, humans vary their admittance continuously depending on the external environment [6].…”

“…The dynamics of the neuromuscular system play an important role when trying to investigate the effect of haptic aids on pilot behavior [2]. Previous works have investigated methods to estimate the neuromuscular system during force-related tasks [3] or more complex control tasks [2], [4], [5]. However, a key limitation with much of the literature is that the estimation methods assume a time-invariant neuromuscular response.…”

Pilot Adaptation to Different Classes of Haptic Aids in Tracking Tasks

“…In a control task with haptic and visual cues, two different dynamic responses need to be investigated: the visual response and the neuromuscular response [20,21]. The visual response describes the pilot response to the visual cue, whereas the neuromuscular response represents the dynamic settings of the pilot's arm and determines how the pilot interacts with forces (from disturbances or from haptic aids).…”