Operability Improvement of Human-Machine Systems by Altering Visual Information: A Case-Study of the Inverted Pendulum

When humans operate machines such as cars and drones based on visual information, it is essential to improve the subjective operational feeling, avoid failures, and reduce the learning time, in other words, to improve operability. In this study, to enhance operability, we focused on visual information and proposed presenting the machine’s altered state rather than its exact state. Here, alteration includes prediction, exaggeration, and noise removal based on the machine’s state, rather than mere image processing. Although some systems are difficult to operate even if the state is accurately displayed, this approach can enhance operability while maintaining human decision-making capabilities in such systems. In this study, with a focus on the prediction among alterations, we constructed a prototype experimental system and demonstrated the feasibility of the proposed concept through an experiment. In the experiment, the subjects played a game with the purpose of keeping an inverted pendulum upright as long as possible.We compared the game score, input amount, and pole inclination between two conditions: one in which the game displayed the current state of the inverted pendulum, and the other in which it displayed the prediction 0.6 seconds later. The experimental results with 35 subjects (32 valid) showed that presenting the altered state tended to improve the game score, reduce the input amount, and reduce the inclination of the pole. This is the first experiment demonstrating that altering visual information can improve the operability of human-machine systems where humans have to perform stabilization tasks.