This paper proposed the novel quartz crystal resonator (QCR) force sensor that incorporates plasma direct bonding technology. Conventional QCR force sensors used a plastic adhesive layer for the bonding of the quartz crystalline substrates. A large creep response was observed due to the plastic strain of this adhesive layer, which made it impossible to secure repeatability in measurement. By incorporating plasma direct bonding, a new type of QCR force sensor was fabricated and creep response decreased to a one-15th in 60 seconds.

This paper presents a design of Si strain sensor to integrate of soft bending microactuator. Bending motion can be detected by integrating a strain sensor into the soft bending microactuator. We focused on Si strain sensor that uses semiconductors with high sensitivity to detect minute movement. However, the allowable strain of Si is small, and the bending motion of the soft actuator may cause Si to break. Therefore, in this study, we proposed a structure in which only the sensing part of the Si strain sensor is made of a thin. In addition, the evaluation results of failure strain showed that flexibility is added by thinning. The result will contribute to the detection of minute forces on delicate objects by soft actuators.

In this paper, we aim to engineer the mechanism of body design using pluripotency, which produces various functional elements from the same cell as in animals. As a first step toward imitating the pluripotent fabrication of animal bodies, we propose a design method for constructing a functional robotic body by extracting various properties from the same material. The material used in this paper is McKibben-type artificial muscle, which has been widely used in the field of robotics research. In this paper, we show that by programming the local properties of a long McKibben-type artificial muscle, each part of the muscle can exhibit various material properties when compressed air is introduced to construct a structure or a part of a robot body.

The animal body contains an extremely wide variety of material properties and structures. By reproducing and understanding these complex body structures, we can expect to be able to make more free and advanced leaps in the structure and manufacturing process of future robots. In this paper, in order to reproduce “adhesion and avulsion” in the animal body, we designed adhesion points by changing the surface texture of the skeleton, and fabricated an artificial joint capsule.

In this paper, we focus on the constraint conditions at the boundary of the skeletal-skin structure with cure inhibition of silicone rubber, and experimentally investigate the effect of the difference in the constraint on the mechanical properties. First, we confirm the effect of cure inhibition on the reduction of adhesion and dynamic friction by friction tests, and we obtain the equivalent friction coefficient. Next, we show that the effect of cure inhibition can be reproduced by setting the contact conditions at the boundary between the rigid and elastic materials in the finite element model. From the analysis results, the mechanical properties of the skeletal-skin structure are verified.

This paper presents a robot exploration algorithm which uses the repelling behaviour of anti-aphrodisiac pheromone mechanism. The algorithm can be used in a multi-robot system in which different robots can efficiently explore different areas of a scene while cooperating with each other through proper pheromone deposition. This eliminates the need of explicitly programming each service robot to serve particular areas of the map. The paths taken by robots are represented as nodes across which pheromones are deposited. Results show that the proposed scheme enables multiple service robots to perform cooperative tasks intelligently without any explicit programming.

This paper proposes a cooperative navigation system for multi-robot systems. The proposed system uses a hitchhiking based approach with a driver and follower (hitchhiking) robot. Our method allows the hitchhiker robot to skip redundant computations in navigation like path planning, localization, obstacle avoidance, and map update by completely relying on the driver robot. This allows the hitchhiker robot, which performs only visual servoing, to save computation while navigating on the common path with the driver robot. The driver robot, in the proposed system performs all the heavy computations in navigation and updates the hitchhiker about the current localized positions and new obstacle positions in the map. Experimental results shows that efficient multi-robot navigation can be performed using the proposed method.

We propose a passage classification method using omnidirectional camera images and object detection by machine learning. In this method, image data of 360 degrees horizontally is firstly acquired using an omnidirectional camera. Next, objects such as passages and doors are detected from the acquired image data using the YOLO detector. Finally, it is determined which passage feature is classified according to the type and position of the detected object. We conducted an experiment to classify passage features from images in order to verify the effectiveness of the proposed method. In addition, the proposed method is applied and used for our scenario-based navigation method. We conduct navigation experiments and verify the effectiveness of the applied method.

The demand for autonomous mobiles is expanding due to the labor shortage caused by the declining birthrate and aging population, as well as the need to improve business efficiency. In a previous study, an outlier processing method for probabilistic position information calculated from multiple sensors was proposed as one of the methods for self-localization , which is an important element of autonomous mobile. In addition, as an application of the previous research, a ’reliability map’ was proposed to visualize the reliability of the localization in each region and to indicate the regions where the localization may fail. In this paper, we test the hypothesis that Robustness of self-localization can be improved by sensor expansion based on a ’reliability map’.

We have proposed an online imitation method for path-following behavior based on end-to-end learning of vision and action. However, the proposed method aims to follow a fixed path and cannot dynamically select a path to make a robot move to a destination. In this study, we add a function to select a path to the method so that the robot can move to an arbitrary destination. We introduce the online imitation learning method with the additional function of selecting a path, and then verify whether the system can select a path by experiments using a simulator.

In this paper, we propose a method of self-positioning by a spherical camera using height information of doors. GPS is a general method of self-positioning. However, it does not work properly in the bad reception environment. There are some methods of self-positioning with a spherical camera for an offline environment. However, they need a complicated map and hard advance preparation. Therefore, we aim to estimate the current location with a simple map and height of doors.

In the path planning of autonomous mobile robots, it is desirable to consider not only the free space but also the condition of the road surface. In this study, we assume that the classification result of the road surface condition can be obtained as a probability distribution, and examine the data structure and update law for holding it on the grid map. The data update rule applies the Bayesian filter. In the simulation, the measurement of a single road surface condition and the measurement of two road surface conditions were performed. The results, the proposed update rule worked effectively, and the execution result of the appropriate route plan was confirmed.

Autonomous wheelchairs are expected to help older people to transport. A previous research about autonomous driving on rough terrain, which is one of the road environments where wheelchairs are used, could not jugde traversability accurately. On the other hand, though rough terrain was not examined in any detail, another research developed a motion planning method based on traversability judgement, considering the vehicle model and road surface that causes fatal accidents during the use of wheelchairs. The purpose of this research is to develop a road surface map generation method for motion planning of autonomous wheelchairs on rough terrain. In this paper, we used visual odometry method for self-localization, and found that road surface map generated has a large error when meandering on rough terrain. By correcting self-localization based on visual odometry method with the use of IMU and vehicle speed, the error of the road surface map reduced.

In this study, we are developing a robotic system that can predict the risk of falling by having a mobile robot autonomously detect the three causes of falling indoors: slipping, stumbling, and falling, and quantitatively and visually evaluate the degree of danger. If this system is put to practical use, it can not only clarify the danger of falling, but also lead to advice on how to eliminate the danger. In order for a mobile robot to detect indoor fall hazards, it must have coverage within the work environment. In this paper, we propose a new coverage algorithm that enables a mobile robot to efficiently and completely traverse the environment in an unknown environment where information such as the size of the indoor area, the size and number of stationary and moving objects are not given, and confirm the effectiveness of the algorithm through operational experiments.

As the demand for autonomous mobile robots has increased in recent years, they are required to operate in a variety of environments, and they need to travel an appropriate path according to the environment. One of the indicators is the road surface condition. In this research, we use RNN to discriminate the road surface condition based on the acceleration time series data of the autonomous mobile robot in the vertical direction. By adding the discriminative information to the environment map, the robot is expected to travel on an appropriate route. We believe that this system can contribute to the operation in various environments.

Path planning is a fundamental problem in mobile robots that optimize the path to determine how the robot reaches its goal. In particular sampling-based methods are popular in robotics. However, the sampling-based method is unsuitable for dynamic and narrow environments, unlike the adaptive method due to its prior planning before moving. Moreover, applying the sampling-based method in real-time and dynamic environments becomes more difficult due to its high computational cost. In this paper, we propose an efficient sampling method for roadmap planners to decrease its computational cost and increase the success rate of path planning. The proposed method applies the artificial potential method and map-decomposition method to roadmap planners such as PRM to achieve the goal.

Recognition of road conditions is an important issue for autonomous mobile robots. When considering the route to be traveled by the robot, it is important to avoid collisions with obstacles, but it is also necessary to select a route that is easy to travel in order to improve the accuracy of the robot. In this paper, we propose a new method for recognizing the road conditions using a camera, which is based on advanced image processing technology. In this paper, we describe a method for recognizing road surface materials based on the laser reflection intensity, which is considered to be robust to ambient light, referring to previous studies, and as an application of this method, we discuss a path planning method for avoiding road surfaces made of specific materials.

The purpose of this research is to verify the construction of a large-scale simulator, which is necessary when verifying an unmanned air traffic control system that manages a large number of UAVs. Therefore, in this study, we connected a number of systems that simulate actual aircraft with ROS to the simulator built with Gazebo, and verified how many simulations could be performed in real time. In the experiment, we ran more than 1 to 15 simulators of UAVs using two PCs, and visualized those simulators with Gazebo. As a result, we found that GPU utilization dominates the smoothness visualization in object visualization.

This paper proposes a multi-robot path planner for intelligently accessing a limited number of charging points distributed on the map. Unlike traditional path planners, which mainly consider the shortest path criterion to generate paths, the proposed planner also considers the remaining battery power of the robots, task priority, and robot’s location in the map. It allocates the most appropriate charging station to the robots, which require recharging. Simulation results show that the proposed planner can reduce trajectory re-planning, and plan efficient paths to the available charging points.

In this study, we propose a route planning method using floor maps such as facility information maps as prior information. The floor map does not provide geometric information because the details are omitted and in some cases the ratio of height and width is inaccurate. Therefore, we extract the connection relations of the passages from the floor map as a node graph, and calculate the shortest path from the specified current location to the destination based on the node graph.We use LiDAR to detect the direction and branching of the corridor in the real environment, and then follow the calculated path.

In this paper, we propose a localization method using Downhill simplex method to superimpose two localization maps. AMCL is capable to localize under environment noise, but take time to converge to ground truth. NDT is capable to converge quickly to ground truth, but is not able to localize under environment noise and the environment where the evaluation values are multimodal. Our experiments show that the proposed method is able to converge to ground truth faster than AMCL and localize under environment noise more accurately than AMCL and NDT.

Biohybrid actuators consisting of skeletal muscle and artificial lattice possess unique features such as self-growth and self-repair functions. As a first step in developing a model-based design for biohybrid actuators, we used the existing muscle contraction model to propose a design method that can achieve appropriate performance for muscle specification. Utilizing the actual skeletal muscle of the toads, muscle contraction force under multiple square waves with different frequencies and amplitudes are used to determine the model parameters by fitting them to the simulated forces. Based on the results, the relationship between the muscle specification and the model parameters is obtained, and we successfully designed muscle specification required to generate specific contraction force based on this model.

Cell rotation is required for three-dimensional observation and attitude control of cells in various fields such as cell biology or medicine. Previously, we proposed a method for three-dimensional rotation of microobject based on vibration-induced flow. Vibration induced flow is a localized flow that is generated around a microstructure when the vibrations are applied to the microstructures. In this paper, we assumed that a microobject is rotating at the center of a vortex flow and constructed a model of horizontal rotation. We also evaluated the validity of the constructed model using microparticles and mouse oocytes. In the future, we will perform design optimization of a microstructures rotations of specific targets by utilizing this model.

In recent years, various cell manipulations are necessary in medicine and cell biology. Previously, we proposed a method of cell manipulations like transportation, rotation, and trapping using the vibration-induced flow. The vibration-induced flow is a local flow that induced around microstructures by applying vibration to a chip with microstructures. In this paper, we analyzed the flow around microstructures which classified into four categories by symmetry. And we succeed to switch the movement of 70 μm particles under different directions of the vibration. In the future, we aim to realize the switch of cell manipulations. In addition to the symmetry of structures, we will analyze the flow that occurs when changing other factors of structures like sharpness and achieve switch of other cell manipulations.

In this research, we propose a micromanipulation system using a haptic device with grip function for intuitive and efficient cell manipulation. In this system, the operator manipulates the micromanipulator and injector connected to a holding pipette using the haptic device. In addition, the system returns force sensations to the operator according to the degree of aspiration/expulsion of the holding pipette and the state of the cells. We verify that the proposed system improves the efficiency of micro-manipulation through subject experiments of micro-manipulation using microbeads.

Cell culture technology is a fundamental technology of biotechnology. In adherent cell culture, cell detachment from culture vessel is an essential process. Although enzyme-based cell detachment is the widely accepted technique, it damages cells. Nowadays, cell detachment technique utilizing ultrasound has been proposed. When ultrasound vibration is applied to the bottom of a culture dish, the acoustic radiation force is given to cells, resulting in cell detachment. For the implementation of cell detachment, the acoustic radiation force on cells should be quantified. Therefore, we developed the device to quantify the acoustic radiation force applied to the cells adhered to a culture dish.

The motivation for this research is to propose a method that facilitates the blood separation process, which is originally performed by a centrifuge, with a chip-type device. The motivation for this research is to propose a method that facilitates the blood separation process, which is originally performed by a centrifuge, with a chip-type device. A separation test between micro-beads and liquid was performed using a suspension of polystyrene micro-beads simulating erythrocytes. Finally, we succeeded in extracting only the liquid component by physically trapping the microbeads using the step of the micro-channel. In this study, we report on the design and separation performance of micro-channels.

This paper proposed a cantilever-type QCR force sensor for force measurement at the time of pipette insertion into the oocyte. QCR force sensor has high sensitivity and high rigidity and is suitable for measuring minute forces. We succeed in measuring force change in the insertion of a sharp pipette into the Xenopus oocyte. There was a difference in the change in force level depend on the diameter of the pipette. When the diameter of the pipette tip was large, we could observe a force increase caused by the contact between the pipette and oocyte and a rapid decrease in force due to penetration of the cell membrane. Quantitative evaluation was investigated.

Arabidopsis is a kind of model plant for studying the mechanism of plant growth and development including photosynthesis and environmental stress response. Lots of studies on Arabidopsis requires the analysis of the root tip. Previously, biologists always cut the root tip and put them in special chemical to isolate the protoplast and analyze the RNA or protein of the cells. But this method has a disadvantage. Protoplast are exposed to enzyme solution long time after they are isolated. This might disturb the original gene expression in the protoplast and affect the result of transcriptome result. In this paper, the author proposed a system combining a microfluidic chip and a micromanipulator with a piezo pump which can be used to isolate and pick up a single cell from the root tip so that the protoplast can be picked up shortly after it is isolated. Thus, it would help to acquire transcriptome data that is close to the original expression. This system would benefit the study of Arabidopsis in the future.

The teleoperation system is very popular for various medical, industrial, and automation applications. During the covid19 pandemic, people understood the importance of the teleoperation system because people could not move from one place to another for travel restrictions. Recently cyborg insects have been a hot research topic for search and rescue operations. However, controlling cyborg insects is not easy because insects are living things and have a biological control system. Only experts can balance the biological and electrical control systems. Otherwise, too much stimulation could kill the insects. This study aimed to develop a cyborg cockroach controlling system for search and rescue operations using the teleoperation technique. The teleoperation system communication link was established between Japan to Bangladesh. These experimental results show that the cyborg insects control accuracy 88% using a teleoperation system. Finally, this study will help control the cyborg insects for search and rescue operations by an expert operator anywhere any the world.

In this report, we evaluated the mechanical properties of liposomes by the micropipette aspiration method for the development of molecular robots. We prepared liposomes with the W/O emulsion method and aspirated them on the micropipette using negative pressure generated by the difference of the water head. We measured the deformation length of a liposome which was aspirated in the micropipette with the microscopic image. The deformation was applied to the viscoelastic model. In the result, we derived Young’s modulus (E₀ = 416 Pa and E_∞ = 57.3 Pa) and viscosity (μ = 120.4 Ns/μm). These parameters were smaller than these of cells. We considered that the result was due to the lack of cytoskeleton which keeps the form of cells of the living things. In the future, we will try to measure more liposomes and derive more accurate parameters.

Our photochemical cell membrane perforation method is new approach for mass cell processing, e.g., cell therapy or regenerative medicine. We are trying to establish a new cell processing system, designed for cell suspension and the suspension is handled within a medical bag like package, so the cell membrane perforation process becomes easy handling and can reduce a risk of contamination. One of the advantages of the system should be simultaneous various materials introduction to cells. In this report, we describe the trial results about the simultaneous introduction of fluorescence marker dye and magnetic nanoparticles into cells using the bag system.

This paper propose real time control method of cyborg insect inspired by biological behaviors. In experiments, we successfully demonstrated navigating an individual insect using IMU data based on machine learning analysis with minimum electrical stimulation.

Microrobots driven by bio-actuators have been attracting extensive attention and research interest. However, they can only work in a liquid environment, greatly restricting their practical applications. In this work, we applied the newly developed biomolecular artificial muscle actuator to break this limitation and designed a microrobot aiming to move at the air-water interface. This microrobot was made up of SU-8 and PDMS and driven by photo-patterned artificial muscles. Its 3D microstructure was origami folded from a double-layer structure machined through lithography, without costly 3D micromachining processes. The results preliminarily demonstrated the feasibility of the fabrication and the moving principle of the proposed microrobot.

Actuators play an important role in lab-on-a-chip systems for a variety of microfluidic tasks. Though rapid progress has been made, the existing devices are still unable to meet the requirements of different applications. In this work, we demonstrated the application of the newly developed biomolecular artificial muscle in microfluidic actuation through two examples, a 3-way micro-valve for micro-flow control and a gated micro-cage for manipulation of micro-objects, which have unique characteristics compared with existing devices. This work provides a new paradigm for the fabrication of microfluidic actuators with low cost, and can contribute to the development of microfluidic devices and accelerate their practical applications.

This paper discusses the functional evaluation of biopolymer interfaces by manipulated liquid-gas transition. An air-jet was used as an inducer to generate liquid-gas transition and applied to the liquid covering on a phospholipid polymer coated surfaces. By switching air-jet injection, the liquid exclusion area and subsequent spontaneous liquid recovery were observed. The recovery behavior was found to vary depending on the composition of biopolymer.

In recent years, cell spheroids have attracted a great deal of attention for their applications in tissue engineering, such as drug efficacy evaluation and organ and blood vessel regeneration. Previously, we proposed a method for generating cell spheroids using vibration-induced flow. However, repeatability of generating cell spheroids was low. Thus, in this paper, we discuss the suitable design of flow pattern induced around micropillar to improve the repeatability of cell spheroids fabrication. By comparing the results of cell aggregation and the flow patterns, we confirmed the swirly flow patterns related to the formation of cell spheroids with high repeatability.

In recent year, aggregations of cells and other small objects are in great demand in tissue engineering and regenerative medicine. Previously, Faraday waves generated on the water surface by applying constant vertical vibration to a chamber filled with liquid have been used to aggregate micro-objects into patterns. However, the phenomenon has not been confirmed or analyzed by applying horizontal vibration. In this paper, we confirmed whether it is possible to generate patterns by applying horizontal vibration to a chamber filled with liquid and conducted pattern generation experiments using chambers of various shapes. In addition, we conducted pattern generation experiments using living cells.

In the field of soft robotics, which focuses on the physical flexibility of biological systems, hydrogels have been applied as a material for soft actuator because of their elasticity and light weight. Conventionally, stereolithography used for three-dimensional modeling of hydrogels have been considered problematic due to their high toxicity to living bodies. We fabricate three-dimensional objects by aggregating microgels made of N-isopropylacrylamide (NIPAm), by relatively diamagnetic assembly. NIPAm is temperature-responsive and is expected to be able to partially shrink the microgel aggregates by laser heating. In this paper, we constructed a laser heating system and measured the temperature near the laser spot during the irradiation. Based on the results, the temperature of the laser spot was also estimated. The temperature of microgel aggregates rose to about 33°C, but we need to improve heating performance because it is not enough heating to shrink microgel aggregates.

Cell therapy treats intractable cancers that are difficult to cure with normal immune function. The functions of cells are modified and then reimplanted into a body to treat diseases. But Cell therapy faces challenges of reliable manufacturing system and affordability. For reliable & low-cost cell therapy, we are developing an automatic system that enables cell screening at the single-cell level based on cell image data. LabVIEW controlled photocuring system is introduced and its impact on processing time and single cells sorting method based upon pixel area is introduced. The automation and continuity of this system made it accordable and reliable to use.

This paper introduces a tip growing mechanism in liquid based on a sol-gel transformation of alginate. A gel film generated by cross-linking provides a flow tube and transmits the pressure at the root to the tip like a hydraulic system. In this paper, we describe the principle and verification of the behavior.

In this study, we designed and assembled a virus sampler using a two-fluid nozzle as the dissolution part, as a new method to dissolve microdroplets containing viruses floating in the air into a liquid. In our design, the two-fluid nozzle sucks up the test gas along with the liquid and mixes the test gas and the liquid. In addition, we used sodium chloride as a pseudo-virus and evaluated collection ratios with and without the two-fluid nozzle. As a result, we confirmed that the collection ratio was improved by 35±2.6 % in the case with the two-fluid nozzle compared to the case without the two-fluid nozzle.

A high refractive index (HRI) glass is a promising material for observing micro-and nano-scale objects under an optical microscope with high spatial resolution. However, HRI glass has not been applied to micro/nanofluidic devices. Typically, fabrication of nanochannels is realized on fused silica glass although the combination of HRI and fused silica glasses is difficult to apply for thermal fusion bonding at high temperature because thermal expansion of HRI is significantly different from that of fused silica. Here, we discuss the bonding capability of HRI glass and fused silica glass with micro- and nano-patterns, at room temperature. Through our room-temperature bonding method, we have achieved to obtain micro and nanochannels formed by two different glass substrates. We have also investigated pressure resistance of the channels and surface properties, surface roughness and surface free energy, of the two glass types.

In this paper, we proposed and evaluated a simple and effective microfluidic device to load mechanical stimuli to a large number of cells to investigate the correlation between gene expression and cell morphological changes. Firstly, we fabricated a microfluidic device using MEMS technology to apply a mechanical stimulus (compressive force) to cells. In particular, we modified the device to be able to collect a large number of cells for gene expression analysis. In our experiments, we achieved the compression of cells and cell nuclei. Furthermore, the deformation of each cell and cell nucleus could be accurately observed and a large number of cells could be collected for gene expression assays. Gene expression analysis of the recovered cells revealed that the induction of several characteristic genes occurred.

In recent years, with the development of the life science field, there has been an increasing demand for the measurement of the properties of cells and tissues. In particular, various techniques have been proposed to measure the response of specific cells to various external stimuli, in order to realize unprecedented analysis. In our laboratory, we are developing a small microhand system that can approach cells with dexterity and speed. Furthermore, we have been developing a new end-effector based on this microhand technology to extend the system to various stimuli such as chemical and temperature stimuli as well as force stimuli. In this paper, we report on the development and evaluation of a micro hand end-effector for multimodal stimulation such as local chemical stimulation as well as force stimulation.

Bio-actuators have been attracting interests for their inherent advantages comparing with traditional actuators, like ability to work in micro-nano size, silence, and low heat emission. Along with the progress in the fabrication and evaluation of the bio-actuators, it was realized that a fixation method was required to stretch the muscle tissue to promote its differentiation and to evaluate the mechanical properties by tensile experiment. However, the traditional fixation methods like glue, jig structure and vacuum have problems such as stress concentration to muscle tissue, and difficult and time-consuming manipulation to muscle tissue. To overcome this difficulty, we fabricated magnetic ends on both sides of the microfiber, therefore the magnetic force could be used to manipulate the muscle fiber, to achieve stretch and evaluation of the mechanical properties of the muscle fiber.

Trapping micro-size targets, including microbeads, cells, and embryos, is necessary for many kinds of research. Fabricating the traditional microchannels for trapping is usually time-wasted, high-cost, and complex. Here, we proposed a variety of reconfigurable 3D-printed molds by which microchannels with trapping functions are fabricated. Microfluidic dynamic is the main mechanism to trap particles. This method can significantly decrease the cost and time of microchannel fabrication and also has the potential to enrich its combination of multiple functions.

Tendon-driven mechanism enables to realize a long-reach robot arm to inspect in industrial plants. However, a large elastic elongation of tendons due to gravitational joint torque makes joint control unstable. In this paper, we proposed a planning method to search for a path that includes only stable postures. The stability of the joint angle was formulated using the balance between tendon elasticity and gravitational torque. As a planning algorithm, Rapidly exploring Random Tree was used. The effectiveness of the proposal was demonstrated by manipulation of 5 kg payload by 10-m-long robotic arm.

This paper describes a system for realizing simultaneous and synchronized collection of air dose rates and measurement locations for efficient dosimetry survey and spatio-temporal dosimetry data logging in nuclear facilities. The prototype system, which is currently under development, mainly consists of a 3D LiDAR-SLAM unit and a dosimeter integrated in a ROS framework. In this paper, we present the configuration of the prototype and the preliminary experimental results of dosimeter position estimation using it.

In order to decommission the Fukushima Daiichi Nuclear Power Plant, we have developed a method to estimate the radiation source distribution accurately by searching the map in a limited time. In the proposed method, multiple robots estimate the radiation source distribution by the coarse-to-fine search. The coarse-to-fine search is to estimate the distribution by coarse search of the map, and then to search finely the high radiation source area based on the estimation. As a result of examining the efficiency of the search by multiple robots, the search time was only 1.5% compared to the case that the map was completely divided equally. As a result of estimating the distribution of radiation source with and without the coarse-to-fine search, the coarse-to-fine search is more accurate.