We are tackling the challenges of "control," which involves guiding various systems, ranging from electromechanical systems like cars, robots, and aircraft, to infrastructure systems like transportation, power, and aviation, and even encompassing domains such as healthcare, human emotions, and individual/group decision-making. Our research spans not only engineering studies to develop control methods for these systems but also scientific investigations to unravel the underlying principles essential for control.
We choose our research subjects based on the interests of our members and engage in exploring new problems and finding innovative solutions.
When you hear the word "control," you may associate it with fully automated control systems, such as those used in autonomous driving, where machines handle everything. However, our research scope extends beyond just automatic control systems. Instead, we envision environments where humans actively participate in system operation, like in assisted driving scenarios. By fostering collaboration between humans and machines, we aim to build truly human-centric systems that prioritize safety, comfort, and sustainability at a higher level than what automated control alone can achieve.
To achieve this goal, we conduct research in two key directions: designing the actions that machines can take towards humans and comprehending humans' interactions with machines. Our focus is on creating systems that embrace human-machine collaboration, surpassing the limits of conventional automatic control and delivering a more beneficial and meaningful experience for people.
Our aim is to enhance user and operator comfort and sustainability while ensuring system safety. We seek to address questions related to what actions machines can take towards humans, how to measure their effects, and how to design systematic approaches to achieve these goals.
One particular aspect we are excited about is "Visual Nudges." By providing visual stimuli to individuals, we can induce behavioral changes through indirect control. For instance, visually alerting a car driver to the presence and danger level of pedestrians may encourage avoidance behaviors, or running pace-maker lights in tunnels can subconsciously prevent speeding. We are applying control theory to address such challenges.
Additionally, we are researching the use of "Incentives" to promote behavioral changes in individuals and groups by providing rewards such as monetary incentives. This has applications in demand control in power systems and traffic congestion mitigation within transportation networks.
Through our research, we aim to strike a balance between system safety and the enhancement of user and operator experiences, employing innovative control approaches such as Visual Nudges and Incentives to achieve our objectives.
Recent Works:
R.Asanaka and M. Inoue, Gig Work Systems Design for Dynamic Work Management with Freelancers, 2023 IEEE International Conference on Systems, Man, and Cybernetics, 2023 (accepted)
T. Udagawa and M. Inoue, Tolling for traffic flow networks: Positive systems modeling and control, IFAC World Congress 2023
M. Takeda, M. Inoue, X. Fang, Y. Minami, and J. M. Maestre, Light guidance control of human: Driver modeling, control system design, and VR experiment, 4th IFAC Workshop on Cyber Physical and Human Systems, 2022 (link)
R. Asanaka, M. Inoue, T. Homma, and K. Sawada, Incentive and nudge design for human behavioral change, SICE Journal of Control, Measurement, and System Integration, 2023
To execute appropriate actions from machines to humans, it is crucial to understand human states, especially intentions and preferences that may not directly manifest in their actions. For instance, we quantify pedestrians' gaze and movements from camera images in vehicles, converting this data into a "system-to-human trust level" to enhance safety in driving assistance. We also conduct research to actively gather feedback from humans, advancing our understanding of them. By utilizing actions from humans to machines, such as "chat" or "surveys," we estimate and model users' intentions and preferences. Based on these models, we redesign the specifications of control systems, enabling the realization of personalized systems tailored to individual users.
Recent Works:
S. Ejaz and M. Inoue, Trust-aware safe control for autonomous navigation: Estimation of system-to-human trust for trust-adaptive control barrier functions, under review.
T. Nii and M. Inoue, Personalized control system via reinforcement learning: Maximizing utility based on user ratings, SICE Journal of Control, Measurement, and System Integration, 2023 (link).
T. Nii and M, Inoue, Personalization of control systems by policy update with improving user trust, IEEE Control Systems Letters, Vol.7, pp.889-894, 2022 (link).
K. Hara, M. Inoue, and J. M. Maestre, Data-driven human modeling: quantifying personal tendency toward laziness, IEEE Control Systems Letters, Vol.5, No.4, pp.1219--1224, 2021 (link)