Fujimoto Lab - Automation & Optimization Lab
Toward the singularity breakthrough by mechatronic intelligence

Our laboratory works on the following research themes to realize a sustainable society.

1. Support and automation technologies for an aging society
• Research on robot intelligence
• Research on robots that support humans with soft mechanisms and control
• Research on high-level automation of construction machinery
• Research on environmental recognition and motion planning using machine learning
2. Technologies that contribute to solving the problem of climate crisis
• Energy-efficient electrical-mechanical energy conversion
• Electrical-mechanical energy conversion based on new principles
• Electrical, mechanical, and thermal energy conversion for climate control

We conduct research primarily in control applications for robots and electromechanical systems, as well as system optimization for production systems and related areas. In particular, in response to pressing societal challenges such as the aging population and climate crisis, we are engaged in the development of robots aimed at supporting the independence of disabled people and reducing the burden on medical and caregiving professionals. Our work also includes research on automatic tuning of control parameters to improve quality and optimal motion planning for various autonomous construction machinery to enhance productivity. Additionally, we are investigating actuators that improve the energy efficiency of electric vehicles and robots, as well as engineering methods to mitigate typhoon impacts and harness energy from typhoons. We actively promote the social implementation of our research seeds and are conducting numerous collaborative studies with both domestic and international companies.

Data-Driven Simulation

Data-driven control attracts attention as it allows for tuning control systems using only a few sets of input-output experimental data. Since this approach does not require deriving a mathematical model of the controlled plant, it simplifies the design of control systems. Our laboratory proposes a new data-driven simulation and controller design method based on convolution operations. This method enables estimation of the system output with only straightforward calculations and minimizes various evaluation functions such as overshoot and settling time. As a result, improving control performance and reducing the effort required for tuning control systems in many industrial applications becomes possible.

Bilateral Drive Gear

Robot arms that may come into contact with humans, such as collaborative robots, must move compliantly in response to external forces from the environment. The ease with which joint actuators respond to such external forces is called backdrivability. Conventional joint actuators that use high-reduction-ratio reducers are generally not backdrivable. Therefore, various methods have been employed to achieve backdrivability, such as combining built-in torque sensors with control schemes, introducing elastic elements, or using high-torque motors paired with low-reduction-ratio reducers. Our laboratory proposes and develops bilateral gears that can achieve high backdrivability and efficiency even at high reduction ratios by optimizing the tooth numbers and profile shift coefficients of compound planetary gear systems.

Autonomous Mobile Robot

Autonomous mobile robots have become very common in service robotics applications in recent years. Current autonomous mobile robots require a pre-built map of the operating environment, which poses challenges for navigation in unknown or dynamic environments. Our laboratory conducts research based on self-localization, map construction, and trajectory planning for autonomous mobile robots using LIDAR. We are working on pedestrian detection and trajectory estimation, path planning using only abstracted simplified map information, and route extraction from map data.

Rehabilitation Robots

We conduct research on various types of rehabilitation robots designed to support humans, including robotic canes, robotic walkers, backpack-type torque generators, powered exoskeletons, and robotic seats. Our work spans a wide range of activities, from the mechanical design and fabrication of original hardware to the development of circuit boards, design and implementation of control algorithms, and system verification. We also study methods for extracting user intent and appropriately reflecting it in control parameters.

Compact Air-cooled Motor

We propose an air-cooled motor to improve torque density, featuring a simplified centrifugal fan structure integrated into the rotor shaft. Air is drawn in through a hole at the end of the rotor shaft, flows through the air gap and winding slots, and is expelled through holes located on the outer periphery of the housing. The simplified centrifugal fan has a compact and straightforward structure, formed by drilling an axial hole along the rotor shaft and several radial through-holes. It enables efficient removal of heat from the windings and permanent magnets, achieving high power and torque densities.

Resonant Motor

We are conducting fundamental research on air-core motors, which eliminate the iron cores from both the stator and rotor of an induction machine consisting solely of coils. Due to the absence of iron cores, the magnetic coupling between the stator and rotor windings is significantly small. However, we can generate torque by inducing large resonant currents through the resonance between capacitors connected to the stator and rotor windings and the winding inductance. We envision this type of motor for use in environments where magnetic materials are not permitted, such as MRI examination rooms.

Helical Motor

There are various linear actuators, including hydraulic/pneumatic actuators, rotary motors with motion conversion mechanisms, and linear motors. In recent years, a trend has been toward replacing hydraulic actuators with electric actuators due to better maintainability. Our laboratory has proposed and developed a unique helical motor with both the mover and stator having a helical structure. The mover performs a helical motion within the stator, which we extract and utilize only the linear motion. Since the air-gap area between the mover and stator can be large, this structure allows for greater thrust compared to conventional linear motors of the same volume, and it also features low mechanical losses due to the absence of mechanical contacts.

Nine-Switch Inverter

A typical three-phase inverter that drives a three-phase load consists of six power devices. This study proposes a nine-switch inverter capable of driving two independent three-phase loads using nine power devices. This configuration reduces the number of power devices to three-fourths. However, the drawback of reduced voltage utilization makes it more suitable for low-voltage applications. Variations such as a twelve-switch inverter, which can drive more loads, also exist. In addition, we are conducting research on multilevel inverters with a reduced count of power devices.