Understanding and mitigating biodynamic feedthrough in human-machine-interfaces

Studying how accelerations cause involuntary control inputs

Inside a moving vehicle humans are subjected to accelerations. These accelerations may interfere with manual control tasks in several ways, reducing manual control performance, control accuracy and - above all - safety during the operation of a vehicle. These phenomena are called biodynamic interferences and are known to occur in a large range of vehicle such as helicopters, aircraft, electric wheelchairs and hydraulic machinery. In my Ph.D. research I am particularly interested in biodynamic feedthrough (BDFT), which describes how vehicle accelerations feed through the human body and cause involuntary limb motions (e.g., involuntary motions of head, arms, legs and hands). These involuntary motions can result in involuntary control inputs.

Research goal

We measure and model biodynamic interferences with the aim to increase our understanding of how accelerations influence manual control behavior. These insights will allow us  to design methods to mitigate biodynamic interference effects. Those methods could, for example, be employed to make flying helicopters safer and controlling electric wheelchairs easier.

Affiliations

This Ph.D. project is a collaboration between the Max-Planck-Institute for Biological Cybernetics and Delft University of Technology.

Within the Max-Planck-Institute for Biological Cybernetics, I am affiliated with the following groups:

• myCopter - Enabling Technologies for Personal Air Transportation Systems
• Cybernetics Approach to Perception and Action

Within Delft University of Technology, I am affiliated with the following groups:

• BioMechatronics & BioRobotics (faculty of Mechanical Engineering)
• Control and Simulation (faculty of Aerospace Engineering)

Understanding and mitigating biodynamic feedthrough in human-machine-interfaces

Inside a moving vehicle humans are subjected to accelerations. These accelerations may interfere with manual control tasks, reducing manual control performance, in several different ways. For example, accelerations may cause involuntary limb motions, which can lead to involuntary control inputs. This phenomenon is called biodynamic feedthrough (BDFT). BDFT is known to reduce ride comfort, control accuracy and - above all - safety during the operation of large range of vehicles (e.g. air- and rotorcraft, automobiles, hydraulic and electric machinery) and can occur under many different circumstances. Particularly, in closed-loop systems, e.g., a pilot flying a helicopter, BDFT induced inputs can develop into a biodynamic coupling: divergent or weakly damped oscillatory involuntary control inputs, which can have catastrophic consequences. Accelerations may also interfere with manual control performance through other mechanisms, such as visual impairment (‘blurring’ of visual input) or neuro-muscular interference (inducing errors in the percieved state of muscles and limbs). The combined effect of accelerations on manual control performance is labelled as biodynamic interference (see Fig. 1).

Fig. 1: Biodynamic interference refers to all mechanism by which accelerations interfere with manual control tasks. In our research we focus on biodynamic feedthrough.

Research Goal

Although biodynamic interferences have been studied in the past decades, many of their underlying mechanisms are only poorly understood. By measuring and modelling these effects we aim to increase our understanding of how accelerations influence manual control behavior. These insights will allow us  to design methods to mitigate biodynamic interference effects. Those methods could, for example, be employed to make flying helicopters safer and controlling electric wheelchairs easier. In our current research we primarily focus on biodynamic feedthrough.

Fig. 2: A model containing all relevant elements in an open-loop biodynamic feedthrough system. The figure was published in Venrooij et al. (August-2011).

Adapting to the human operator

The occurrence of biodynamic interference is influenced by many different factors, which often show complex interactions. Examples of such factors are the control device dynamics and the acceleration frequency characteristics, but also more elusive concepts such as mental workload and fatigue. A particularly challenging aspect in biodynamic interference is the role of the human operator. Our research has shown that the limb dynamics of the human body influence the biodynamic feedthrough dynamics (see Fig. 3). Limb dynamics do not only differ from person to person, but can also vary significantly over time. Limb dynamics depend, amongst other factors,  on the manual control task the human operator is performing. For some tasks a ‘stiffer’ setting of the neuromuscular system may be beneficial where for other tasks a more ‘compliant’ setting is required. In order to mitigate biodynamic interference effects for varying settings of the limb dynamics, a cancellation solution is needed that is capable of adapting to changes in the state of the human operator.

Fig. 3: The magnitude of biodynamic feedthrough for different control tasks (each requiring different limb dynamics).

Experimental research into biodynamic interference

The research efforts into biodynamic feedthrough can be subdivided in three main pillars: measuring BDFT, modeling BDFT, and cancelling BDFT. Measuring BDFT is done in experiments, using MPI's CyberMotion Simulator (Fig. 4) and SIMONA Research Simulator of Delft University of Technology. A method was developed to experimentally measure the relationship between biodynamic feedthrough and neuromuscular admittance (i.e., limb dynamics) using actuated control devices (Fig. 4). In these experiments, subjects are requested to perform manual control tasks while being subjected to both acceleration and force disturbances. To investigate the effect of limb dynamics on BDFT effects, the control tasks are designed such that a subject has to attain different settings of his/her neuromuscular system, resulting in different limb dynamics for each task, ranging from ‘stiff’ dynamics (in the ‘position task’) to ‘compliant’ dynamics (in the ‘force task’). From the experimental results, models describing the BDFT effects are developed. Finally, using these models, ways to mitigate BDFT effects are proposed and investigated.

Another experimental study investigates the effect of acceleration perturbation on the accuracy of reaching tasks. The results of the experiment will provide insight in how accelerations may impede the use of touch screen interfaces in acceleration envorinments.

Fig. 4: The MPI CyberMotion simulator (left) and MPI’s actuated control devices (right)

Joost Venrooij

Joost Venrooij is a Ph.D. student in a collaborative project between the Max-Planck-Institute for Biological Cybernetics and Delft University of Technology. His interests include biodynamic interferences, neuromuscular modeling and haptics.

Education

Experience

Honors and awards