Force Interaction with a Small Insect by a Soft Micronger with Active Tactile Sensing

Human-robot interaction technology has contributed to improving sociality for humanoid robots. At scales far from human scales, a microrobot can interact with an environment in a small world. Microsensors have been applied to measurement of forces by ying or walking insects. Meanwhile, most previous works focused on the measurement of the behavior of insects. Here, we propose microrobot-insect interactions by soft microngers integrated with articial muscle actuators and tactile sensors, which has been developed for a haptic teleoperation robot system. A soft pneumatic balloon actuator acts as the articial muscle, and a exible strain sensor using a liquid metal provides tactile sensing. Force interaction between a pill bug and the micronger could be accomplished. The microger (12 mm × 3 mm × 490 µm) can move and touch an insect, and it can detect reaction force from an insect. The measured reaction force from the legs of a pill bug as a representative insect was less than 10 mN. Our results demonstrate the potential of micronger-insect interactions in a small world. We anticipate that our results will lead to novel human-environment interaction.

Downsized microsensors are suitable to functionalize robots without disrupting their fundamental functions. Furthermore, micromachines have potential as an intermediary tool for various interactions with a small world. Microrobots are capable of interacting with an environment in this small world, whereas humanoid robots are designed for human-robot interaction in a macro world. A combination of haptic interfaces with microrobots could even allow interaction between the small world and us.
Microsensors have been used for force measurement of small living things such as insects. The ight force of ying insects, as a typical force of insects, has been measured by various means [1][2][3][4][5][6]. Direct measurement by microsensors and image processing for motion capture have been used for force measurement. The deformation, motion and generated force of the wings of a moth were optically measured using fringe pattern projection [1]. The aerodynamic vertical force was approximately 7 mN, which is approximately 5 times stronger than the gravitational force acting on the moth (approximately 1.3 mN).
The ight forces of Drosophila were measured using a MEMS capacitive force sensor (3.6 mm × 2.1 mm × 0.5 mm) to understand ight biomechanics in Drosophila (3 mm long) [2]. The capacitive sensor was developed using a silicon on insulator (SOI) substrate to capture instantaneous ight force in real time.
Drosophila samples were tethered to a sensor probe (3 mm × 50 µm × 50 µm). The total ight force was estimated at a few tens of micronewtons. The collision avoidance behavior of the locust (40 mm in body length) was investigated with simultaneous force measurements and high-speed video recording.
Interesting results on the relationship between wing apping, lift and thrust were reported [3]. The force and moment of the takeoff ight of fruit ies were analyzed using high-speed video techniques [4]. The contribution of the force of the jumping legs and the apping wings to y lifting was examined. The vertical force of the jumping legs (µN order) is su ciently larger than the corresponding aerodynamic force. Social forces in the interaction by laboratory swarms of the ying midge Chironomus riparius were studied using multicamera stereoimaging and particle-tracking techniques to understand collective animal behavior [5]. The acceleration from each insect toward its nearest neighbor was measured to estimate repulsive and attractive forces in that study. The traction force by male Strepsiptera (Insecta) was measured to estimate the dependence of the surface condition of substrates [6]. The force was measured using a force sensor based on strain gauges, which were attached to the insects through a thin polymer thread. The mean values of the measured force were lower than 0.5 mN.
In addition to ight force measurement, leg forces of various insects were measured [7][8][9][10]. Measurement of the force of the leg of a walking stick insect was reported to investigate the control mechanism in positioning the joint in the legs [7]. A platform with a forcemeter was prepared in the path where stick insects walk. The median of difference between the force value at the beginning and end of the stimulus ramp was -3.0 mN ( exion) and 6.0 mN (extension). A multiaxis piezoresistive sensor with micronewton force resolution was reported to measure the foot force of insects such as ants smaller than cockroaches [8]. The demonstrated sensor had a minimum force resolution on the order of 0.5 mN. The plant-insect interactions on lea ets were investigated by measuring attachment (traction) forces generated by beetles on various plant substrates [9]. The dorsal surface of the beetle thorax was attached to a load cell force sensor by means of hair. The measured maximal traction forces on plant surfaces differed from the force generated on glass as the control ranging from 0.5 to 11.8 mN. An array of micro force plates using strain gauges for the measurement of ground reaction forces of insect legs was reported [10]. The force resolution was 1 µN. Eight force plates were arrayed to measure the ground reaction forces of an ant.
Not only animals but also plants generate forces to deform their shape and alter their physical characteristics. The motion of the Venus ytrap upper leaf is an example of a well-known motion generated by a plant. Measurement of forces generated by the Venus ytrap, which strikes, holds and compresses the prey, was reported [11]. A piezoelectric sensor was used for direct measurements of the average impact force of the trap together with a video camera for the determination of time constants.
The impact average force between rims of two lobes in the Venus ytrap was found to be 149 mN, for example.
Most previous works focused on the measurement of the behavior of insects, such as ight forces and leg forces. This paper, for the rst time, presents microrobot-insect interactions by a soft micro nger integrated with an arti cial muscle actuator and tactile strain sensor as shown in Fig. 1(a). A micro nger can apply force to an objective insect and stimulate the insect. The arti cial muscle actuator for a micro nger, which is a pneumatic balloon actuator (PBA) made of polymer, is soft and safe enough to interact with insects gently [12,13]. Manipulation robot systems with micro ngers have been developed using PBA for object-grasping motion [12]. A minuscule sh roe was successfully manipulated.
Furthermore, micro ngers for cellular aggregate manipulation have been developed [13]. A spherical human mesenchymal stem cell (hMSC) aggregate (φ200 µm) was pinched and released on a microwell plate. Our study, in addition to arti cial muscle microactuators, incorporates the integration of tactile sensors into a micro nger. A exile tmeperature sensor was integrated into a micro nger for temperature sensing functionality [14]. Several types of strain sensors were studied for motion detection of a micro nger. Recently, a strain sensor using a microchannel lled with liquid metal (Galinstan) was developed for PBA [15]. The micro nger can detect the reaction force form an insect. Therefore, the micro nger enables active force sensing against living insects. Furthermore, it is possible to transmit the detected force to an operator. We have developed and reported a haptic teleoperation robot system composed of a slave micro nger and a master interface device for an operator [16,17]. The system is capable of presenting touch sense detected by the micro nger to the operator. This paper presents the force interaction between a pill bug and a micro nger, which shows the potential of micro nger-insect interactions in the small world.
Results And Discussion Fig.1 (a) illustrates an image of micro nger-insect interactions. A pill bug is illustrated as a representative insect in Fig. 1 (a). Micro ngers move and apply force to a pill bug lying on its back. The pill bug reacts to the force and pushes back on the micro ngers. The micro ngers can detect the reaction force from the pill bug by their strain sensors. Active sensing by the micro ngers enables micro nger-insect interactions. Fig. 1 (b) shows a photograph of a developed microhand with ve micro ngers. This study focuses on a single micro nger integrated with arti cial muscle microactuators (PBAs) and a tactile sensor.
Micro ngers for active force sensing of a pill bug respectively. The developed micro nger was 12 mm long, 3 mm wide, and 490 μm thick, and was designed for a pill bug. The body length of the sampled pill bugs ranged from 8 mm to 16 mm. The micro nger was made of polydimethylsiloxane (PDMS) and integrated with a PBA and a strain sensor using liquid metal [15].

Figs. 3 (a)-(c)
show the characterization results of the bending motion of a micro nger and the force generated by a micro nger. The relationship between the applied pressure and bending angle is shown in Fig. 3 (a). The bending angle increased in accordance with the applied at pressure at pressures higher than 70 kPa. Fig. 3 (b) summarizes the output characteristics of a micro nger in terms of both bending angle and generated force in accordance with the applied pressure. Fig. 3 (c) shows the characteristics of a exible strain sensor for a micro nger. A strain sensor used the electrical resistance of liquid metal.
Galinstan (eutectic gallium indium stannum) as a typical liquid metal was lled into a microchannel for a exible strain sensor with good compatibility with PBA using microchannels for pressure supply. A strain sensor showed linear characteristics against the bending angle.
This study aimed to measure both leg force and abdominal force of a pill bug. The applied pressure for force measurement experiments was set at 140 kPa, which could generate 15.5 mN. Figs. 4 (a) and (b) show photographs of active sensing against a pill bug lying on its back. Supplemental Videos S1 and S2 show more details of the measurement experiments, especially the motion of a pill bug. Active sensing by a micro nger was used to measure the forces of a pill bug. A pill bug was immobilized by applying suction to its back. A vacuum tweezer device was arranged and used as an insect xing base. Fig. 4 (a) shows active sensing of a leg force of a pill bug, whereas Fig. 4 (b) shows active sensing of an abdominal force. The attitude angle of a pill bug was adjusted according to requirements for optimum contacts against the leg and body, as shown in Fig. 4. The angular position of a suction tool xing the back of a pill bug was controlled. A pill bag was set at sideward posture in the leg force measurement, whereas it was adjusted at upward posture in the abdominal force measurement.
Force measurement of a pill bug by a micro nger  Fig. 5 (a). The leg motion happened at approximately 0.3 Hz. According to supplemental Video S1, detected reaction showed higher frequency (1.6 Hz) because a plurality of legs pushed the micro nger alternately. Measured abdominal force was larger than leg force and exceeded 10 mN as shown in Fig. 5 (b). The frequency of abdominal motion in Fig. 5 (b) was lower than that of leg motion and was estimated to be approximately 0.03 Hz. The frequency of leg motion was much higher than that of abdominal motion. Supplemental Videos S1 and S2 show differences in leg and abdominal motions. A larger displacement amount of motion pushing back a micro nger tended to show a larger force.
The detected information can be utilized for the haptic teleoperation robot system as presented in Fig. 1 (b), which uses the micro nger as a slave component in combination with a master interface device for an operator [16,17]. The system is capable of presenting touch sense acquired by the micro nger-insect interaction thorough the haptic teleoperation robot system.

Dependence of force and weight of a pill bug
Forces by pill bugs were estimated from the point of view of weight dependence. Thirteen pill bugs were collected for the evaluation. Fig. 6 reports leg forces and abdominal forces of pill bugs in relation to their weight. The weight of the pill bugs was also measured during the force measurement. The weight distribution of the collected pill bugs was between 100 mg and 200 mg.
The abdominal force of pill bugs weighing under 160 mg tended to increase depending on the weight of the pill bug. A pill bug of 158 mg in weight generated a maximum abdominal force of 15 mN according to a linear relationship. However, abdominal force generated by pill bugs whose weight was over 160 mg showed different tendencies. Regarding the sampled pill bugs in our experiments, pill bugs weighing over 160 mg did not show a larger force than those weighing under 160 mg, contrary to the expected linear relationship between weight and force. The tendency of increasing the force was not as prominent in the case of leg force. The leg force was less than 10 mN for the collected pill bugs.

Materials And Methods
Fabrication of a micro nger integrated with PBAs and a strain sensor A micro nger (12 mm × 3 mm × 490 μm) was integrated with PBAs and a strain sensor. Microchannels for both PBA and a liquid metal-based sensor were fabricated by molding PDMS (Silpot 184, Dow Corning Inc.) on a photoresist (SU-8) mold on a Si substrate. Bonded PDMS lms were equipped with PDMS interconnections. Galinstan (eutectic gallium indium stannum, Zairyo-ya.com) was injected into microchannels for a sensor and wired through an interconnection.
Derivation method for reaction force from an insect  Fig. 7 (a). The bending angle of a micro nger becomes θ 1 when a pill bug pushes back a micro nger ( Fig. 7 (b)). (c), respectively. First, the expected force corresponding to the applied pressure is obtained through the obtained relationship in advance (Fig. 7 (c)). We set applied pressure at 140 kPa in our experiments. Next, the current bending angle θ 1 is evaluated using the signal detected by a strain sensor in a micro nger ( Fig. 7 (d)). The relative resistance change ΔR/R0 is measured and converted into bending angle θ.
Finally, a reaction force is calculated using a difference of corresponding forces to θ 0 and θ 1 , as shown in Fig. 7 (e). As a result of the calculation steps from (c) to (e), the relationship between pushing buck angle θ pb (θ pb = θ 0 -θ 1 ) and reaction force F can be obtained as shown in Fig. 8, which is available for the touch sense presentation.

Experimental setup
A suction tool was used to x a pill bug by sucking its back. A vacuum tweezer device (TWEEZER-VAC TM , VIRTUAL) was utilized as an insect xing base. The pill bug was immobilized by a suction pump (ASPIRATOR Labo Helper, TOKYO M.I. COMPANY, INC.). The electrical resistance of the liquid metal-based strain sensor was measured by a digital multimeter (34460A, Keysight Tech.) in combination with data processing software (BenchVue).

Declaration Data availability
All data generated or analyzed during this study are included in this published article. Figure 1 Micro nger-insect interaction. (a) Schematic drawing of micro nger-insect (pill bug) interaction. (b) Photograph of a developed microhand with ve micro ngers. This study focuses on a single micro nger, whereas the microhand in the photograph (b) implies the potential of human hand-insect interactions through the haptic teleoperation robot system.   Force measurement of a pill bug by a micro nger. (a) Active sensing of leg force. (b) Active sensing of abdominal force. The pill bug was immobilized by vacuum tweezer device. The attitude angle of a pill bug was adjusted for optimum contacts against the leg and body. The generated force by a micro nger was 15.6 mN at 140 kPa.