Abstract. This paper proposes a conceptual approach to constructing combined feedback in a human–machine interaction system through introducing an artificial sensory feedback component controlled by a technical subsystem. The approach is intended to systematize the role of combined feedback in the control of multi-agent systems with additional elements, humans, and artificial agents. This approach is studied for human vertical posture control and in synthetic experiments (within the CartPole model) considered using reinforcement learning as an example. The efficiency of the control problem solution is investigated by varying the characteristics of information transmission channels and the properties of the artificial sensory feedback component. According to the results, natural experiment observations are conceptually similar to those of the artificial numerical experiment in terms of additional feedback channel operation: there are a similar overshoot effect and prospects for improving control performance by tuning the artificial sensory component.
Keywords: human–machine interaction, optimal control, feedback, reinforcement learning, multi-agent systems.
Acknowledgments. This work was supported by the Russian Science Foundation, project no. 24-11-00272.
An Artificial Sensory Component in a Man–Machine System with Combined Feedback