2.1. System Configuration and Drive Mechanism of the Edible Mobile Robot

The mobile robot proposed in this study employed a four-wheel-drive (4WD) system to ensure reliable propulsion. Fig. 2 shows an overview of the edible mobile robot featuring a hybrid structure comprising edible and non-edible components. The tires in contact with the road surface, gears and timing belt for the power transmission, and axles supporting the vehicle body were fabricated from edible materials. Meanwhile, the body, which serves as the structural foundation, was fabricated from 3D-printed polylactic acid (PLA), with external dimensions of 114 mm (length), 60 mm (width), and 40 mm (height). Additionally, a low-speed, high-torque motor (operating voltage: 3 V, rotational speed: 15 rpm) was utilized as the power source. As the drive principle, the output power from the motor was initially transmitted to an edible gear fixed to one of the axles. Furthermore, the gears mounted on both the front and rear axles were connected via a timing belt made of an edible material, thereby distributing the driving force of the motor to all the wheels and enabling the synchronous rotation of the four wheels.

2.2. Material Selection and Component Fabrication Processes

The components of the robot were manufactured using materials and processes selected according to the mechanical requirements. First, the gears and timing belt for power transmission were fabricated by pouring a gummy solution into silicone molds, and subsequently cooled and solidified. This gummy solution was formulated using gelatin (Cook Gelatin, Morinaga & Co., Ltd., Japan), glycerin (Vegetable Glycerin, Amazing Craft Co., Ltd., Japan), water, and citric acid (Food Additive Citric Acid, Kenei Pharmaceutical Co., Ltd., Japan) at a mass ratio of \(2.0 : 1.5 : 2.5 : 0.2\).

Specifically, a gear train with a reduction ratio of 1.2, consisting of a 12-tooth driven gear (blue, pitch circle diameter: 24 mm) and a 10-tooth driving gear (yellow, pitch circle diameter: 20 mm), based on the standard Module 2 specifications, was arranged with a center distance of 100 mm (Fig. 3(a)). Additionally, a timing belt with 40 teeth, width of 18 mm, and circumference of approximately 269.2 mm was fabricated (Fig. 3(b)).

The axles, which support the weight of the robot and are subjected to torsional stress, require higher mechanical strength; therefore, a hard candy (“Okuchi Sukkiri Washu Hakka Nodo Ame,” Asahi Group Foods, Ltd., Japan) was selected. The candy was melted using a microwave oven (500 W) for 90 to 120 seconds, poured into an axle mold, and then cooled to solidify.

As the tires directly contact the road surface and support the loads and impact, additive manufacturing was applied via a 3DFP, using cookie dough as food ink. After 3D printing, the fabricated wheels were baked in an oven at 150 °C for 12 minutes. The 3DFP allows fabricating fine structures and complex internal geometries that are difficult to mold using conventional casting methods. Moreover, as a wide variety of wheel geometries can be fabricated by modifying the parameters in the 3D computer-aided design model, this approach facilitates optimizing the wheel morphology for different road surface conditions.

2.3. Properties and Formulation of the Food Ink in 3D Food Printing

The properties of the food ink must be highly compatible with the screw-extrusion 3DFP to fabricate the complex internal structures of the tires (discussed in Section 2.2) accurately and obtain a durable and practically usable product. Specifically, the ink should exhibit a stable fluidity suitable for extruding from the nozzle during the printing process, and transforming into a solid, with sufficient mechanical strength to support the vehicle body, after heating. These irreversible physical changes induced by heating enhance the adhesion between the deposited ink layers, ultimately determining the durability of the final printed object. In this study, a rice flour-based cookie dough was selected as the material due to its excellent dimensional stability and shape retention subsequent to heating. Furthermore, the optimal rheological properties that achieve both the extrusion stability in 3DFP and mechanical strength of the printed object were explored by varying the amount of the base rice flour at three levels: 60 g, 65 g, and 70 g. For each portion of rice flour, 30 g of sugar, 30 g of rice oil, and one whole egg were added, and the ingredients were uniformly kneaded using a mixer. This standardized formulation and mixing procedure ensured preparing the food ink with high reproducibility.

2.4. Internal Structure of the Edible Wheels

Four types of wheels, i.e., (a) spoke, (b) mesh, (c) fin, and (d) dish (Fig. 2) were designed for fabrication using the prepared food ink.

1. (a)

   The spoke wheel features a structure connecting the central hub and the outer rim with radial members, which ensures lightweight characteristics and shock absorption through moderate flexibility.

2. (b)

   The mesh wheel consists of mesh-like members arranged between the hub and the rim. This design alleviates the localized stress caused by external forces and distributes the loads along multiple directions.

3. (c)

   The fin wheel incorporates diagonal or curved blade-like protrusions relative to the direction of travel, imparting specific bending stiffness and stress transmission in the rotational direction.

4. (d)

   The dish wheel features a structure wherein the central hub is substantially expanded toward the outer periphery, with thick supporting members reinforcing the rim.

Fabricating these four designs using 3DFP and measuring their mechanical strengths via texture analysis enabled quantitatively evaluating the impact of differences in the internal structures on the durability of the edible wheels.

3.1. Evaluating the Rheological Properties of the Food Ink

The rheological properties of the rice flour-based food ink were measured using an MCR302 rheometer (Anton Paar, Austria) to evaluate its printability. Typically, these properties significantly affect the extrusion stability and printing accuracy in screw-extrusion 3DFP. A parallel plate with a diameter of 25 mm and a gap height of 1 mm was used for the measurements, and cookie doughs with three levels of rice flour content (i.e., 60 g, 65 g, and 70 g) were evaluated.

First, a strain sweep was conducted as a dynamic viscoelastic measurement to determine the linear viscoelastic region (LVR) over a strain range of 0.01% to 150% under steady conditions, at a temperature of 25 °C and a frequency of 1 Hz. Thus, the storage modulus (\(G'\)) and loss modulus (\(G''\)) were measured, and the structural integrity and energy dissipation characteristics of the ink were evaluated to verify its capability for structural recovery after applying stress. Furthermore, a frequency sweep measurement was performed in the range of 0.1 Hz to 100 Hz to calculate the apparent viscosity. These analyses revealed the correlation between the microstructure of the ink and its recovery upon releasing the stress, providing crucial insights for evaluating its suitability for the 3D printing process.

3.2. Verification of the Printing Accuracy of the 3DFP

Printing experiments using a 3DFP were conducted, based on the rheological evaluation, to compare the differences in the printing accuracy between the various ink formulations. The 3DFP (FP-3000, Seiki Co., Ltd., Japan) used in this experiment is shown in Fig. 4. This machine features a mechanism that continuously extrudes the material from a syringe via a motor-driven screw, rendering it highly suitable for the stable additive manufacturing of highly viscous paste-like foods. Although the printer is equipped with a dual-nozzle mechanism that enables two-color printing, only a single nozzle was utilized in this study. The printing speed was 5000 mm/s, and the nozzle diameter was 1.0 mm. A cube measuring \(\textrm{25~mm} \times \textrm{25~mm} \times \textrm{25~mm}\) was printed to assess the shape retention immediately after printing and after heating.

3.3. Mechanical Strength Evaluation of the Edible Wheels and Axles

The ink formulation that exhibited the highest shape retention in the printing accuracy evaluation (Section 3.2) was adopted as the material for fabricating the tires. The wheels featuring the four types of internal structures were all fabricated with a thickness of 5 mm using 3DFP. In addition, the axles were fabricated via mold-casting using candy. Fracture tests were conducted using a creep meter (RE2-33005C, Yamaden Co., Ltd., Japan) to evaluate the durability of the fabricated edible wheels and axles quantitatively. The experimental setup of the fracture test is shown in Fig. 5.

The 3D-printed edible wheels were cut in half along the plane passing through the central axis to prepare semi-circular specimens to observe the deformation and fracture behavior of the internal structures accurately and, thus, evaluate the tires. During the testing, each specimen was placed on the testing stage with the flat cut surface facing downward, and a compressive load was applied radially to the outer circumferential surface of the wheel toward the center. By contrast, the load was applied perpendicularly to the longitudinal axis of the cylindrical axle specimens (diameter: 5 mm, length: 40 mm) fabricated via molding. A wedge-shaped plunger with tip dimensions of \(\textrm{1~mm} \times \textrm{30~mm}\) was employed as the loading jig, and compressed using a 200 N load cell at a constant speed of 10 mm/s until it fractured. Three trials were conducted for each specimen to ensure data reliability and reproducibility. The obtained mechanical responses were plotted as load-strain curves, and the average fracture loads resulting from the different internal structures were compared and evaluated.

3.4. Driving Test of the Edible Mobile Robot

A prototype of a 4WD edible mobile robot was constructed by assembling the gummy gears and timing belt fabricated using silicone molds, candy axles, and 3D-printed cookie tires onto a PLA body. Furthermore, the assembly of the robot and the driving tests were conducted immediately after demolding the edible components to minimize the effects of moisture evaporation and shrinkage over time in edible materials such as the gummy.

Based on the design specifications of the installed motor (driving voltage: 3 V, no-load rotational speed: 15 rpm) and the tire diameter (56 mm), the theoretical ideal propulsion speed was calculated assuming no losses such as slippage in the power transmission mechanism. The theoretical ideal propulsion speed, \(V_{\mathrm{ideal}}\) [mm/s], was calculated using the following equation, with the motor rotational speed \(N_m\) [rpm], the gear ratio \(R\), and the outer tire diameter \(D\) [mm]:

The driving behavior of the robot was video-recorded to evaluate its propulsion quantitatively, and was analyzed using an image processing software (ImageJ). Thus, the displacement was tracked over time using the trajectory of the robot, and the average moving speed was calculated.

4.1. Rheological Properties of the Food Ink

The dynamic viscoelasticity and apparent viscosity measurements are shown in Fig. 6. From the strain sweep measurements shown in Fig. 6(a), the limit of the LVR of the ink was confirmed to be at a strain of less than 1%. Within the LVR, the storage modulus (\(G'\)) exceeds the loss modulus (\(G''\)), and elastic behavior dominates, because the strain is sufficiently small and does not disrupt the physical structure of the material. However, beyond the LVR, the deformation of the material induces a significant decrease in both \(G'\) and \(G''\); when \(G''\) exceeds \(G'\), the ink exhibits a liquid-like flow behavior. Among the three levels compared, the ink formulated with 70 g of rice flour exhibited the highest \(G'\) and \(G''\), indicating that it possessed a robust network structure.

Furthermore, in the apparent viscosity measurements shown in Fig. 6(b), a thixotropic property, wherein the viscosity exhibits an exponential decay, was confirmed in all the formulations. The ink with 70 g of rice flour demonstrated a clear shear-thinning behavior despite having a high initial viscosity. This observation indicates that the ink fluidizes under the shear stress exerted by the 3DFP screw, enabling smooth extrusion, and recovers its original solid-like structure once the stress is released (after passing through the nozzle). In other words, this demonstrates that the rotation of the screw in the 3DFP effectively reduces the viscosity, thereby enabling the extrusion of the ink.

4.2. Verification of Printing Accuracy Using a 3D Food Printer

The impact of the differences in the rheological properties on the actual printing accuracy is shown in Fig. 7. Due to its high storage modulus (\(G'\)), the ink formulated with 70 g of rice flour could support its own weight and maintained the shape of a \(\textrm{25~mm} \times \textrm{25~mm} \times \textrm{25~mm}\) cube even after printing. By contrast, the inks with 60 g and 65 g of rice flour had relatively lower viscosities and moduli, and slumped immediately after printing, which made retaining the 3D complex shapes difficult. Optimizing the post-printing shape retention and extrusion characteristics is essential to achieve the necessary printing accuracy. These results demonstrated that the formulation using 70 g of rice flour is the optimal material for fabricating the targeted internal structures of the wheels.

4.3. Mechanical Strength Evaluation of Edible Wheels and Axles

Compressive fracture tests were conducted on the four types of edible tires fabricated via 3DFP (containing 70 g of rice flour) and mold-cast candy axles to evaluate the mechanical strength of the edible parts constituting the robot quantitatively. The average load-strain curves from three trials for each sample are shown in Fig. 8. Additionally, the statistical data regarding the mass and maximum fracture load from the three trials (\({n=3}\)) for each sample are summarized in Tables 1 and 2.

In the wheel tests, all samples reached their yield points at a strain of less than 10%. Comparing the maximum loads of the average curves shows that the dish wheel exhibited the highest value among the four types at 20.0 N. The dish wheel features a structure wherein the central hub is significantly expanded toward the outer periphery, and the members supporting the rim are thickened. The measurement results (Table 1) suggest that it has the maximum mass (16.6 g), and a high volumetric fill ratio for its shape. This dense material distribution ensures a large effective cross-sectional area, which is considered the primary factor underpinning its extremely high rigidity.

By contrast, the spoke wheel recorded the lowest maximum load at 14.9 N. A structure extending linearly from the hub to the rim, such as that of the spoke wheel, tends to concentrate compressive loads on specific struts. The spoke wheel likely fractured abruptly because its structure was unable to effectively distribute the applied load.

Furthermore, the mesh wheel and the fin wheel exhibited maximum loads of 16.3 N and 15.9 N, respectively. Compared to the spoke wheel, a more gradual load decrease leading up to the fracture was observed in these models. The structures of the mesh and fin wheels likely avoided direct buckling under compression and dispersed the load through the bending deformation of the members, thereby contributing to the gradual fracture behavior.

Moreover, in terms of the maximum fracture load relative to the mass of each shape, the lightest fin wheel (14.8 g) demonstrated higher strength than the heavier spoke wheel (15.1 g). This suggests that the diagonal structure of the fins can support and disperse compressive loads more efficiently, with less material volume, compared to a simple linear radial structure.

The measured fracture loads were compared with the estimated service load to assess whether the fabricated wheels possessed sufficient mechanical strength for practical operation. The total mass of the prototype was 120 g (1.18 N); assuming it is supported equally by the four wheels, the static load applied to each wheel is approximately 0.29 N. However, when operating in actual environments, dynamic and impact loads act on the wheels due to road surface irregularities and acceleration or deceleration. Furthermore, variations in the payload must be anticipated for future applications, such as carrying nutrients or sensors. Therefore, accounting for these fluctuating loads is necessary when designing the wheels. The maximum fracture loads of each wheel measured in this test (14.9–20.0 N) significantly exceeded this static load (0.29 N). Consequently, the fabricated wheels demonstrate a sufficient safety margin for practical operation even considering the dynamic and fluctuating loads.

Next, the stress on the candy axle increased linearly until reaching a maximum load of 26.7 N at a strain of approximately 5%–7%, exhibiting typical brittle fracture behavior where the load dropped sharply at the moment a crack was initiated. As is evident from the comparison in Fig. 8, the candy axle possesses higher rigidity than the cookie wheels, confirming that it has sufficient load-bearing capacity to support the weight of the robot.

However, the brittle nature quantitatively demonstrated in this test suggests a risk of sudden fracture when an impact load exceeding its limit is applied—such as encountering a bump during locomotion—or when the stress is concentrated on the microscopic surface flaws generated during molding. Therefore, adopting a structure that encapsulates a flexible material, such as gummy, as a core is considered effective to enhance the durability of future edible mobile robots. This approach would impart toughness and suppress brittle fracture while maintaining the high rigidity of the candy.

4.4. Driving Test of the Edible Mobile Robot

The propulsion test of the assembled edible mobile robot is shown in Fig. 9. The preliminary evaluations conducted for selecting the tires for this experiment confirmed that, under the driving conditions of the robot and on a flat road surface, the differences in the internal structure of the tire did not result in any significant difference in the straight-line propulsion speed. Therefore, a representative shape with average weight and mechanical strength characteristics was adopted for the propulsion test to ensure the validity of the experimental model and the reliability of the results, instead of selecting a shape that excelled only in a specific performance metric.

Based on Tables 1 and 2, the overall average mass and maximum fracture load for the four types of wheels are 15.5 g and 16.8 N, respectively. The measurement results for the mesh wheel (mass: 15.6 g, maximum load: 16.32 N) indicate that it possessed the average characteristics among the four designed edible wheels in terms of both weight and mechanical strength. Furthermore, the mesh wheel exhibited the smallest variation in fracture load (standard deviation: \({\pm}1.85\) N) among the four types, demonstrating superior structural stability. For these reasons, the mesh wheel was adopted as the representative shape for the propulsion tests.

Per the propulsion test, the output power from the motor was successfully transmitted to the front and rear axles via the gummy gears and the timing belt, confirming continued propulsion using the four-wheel drive system.

In this experiment, the robot traveled a distance of \(\Delta x = 126\) mm over a duration of \(\Delta t = 4.0\) s. The measured average propulsion speed \(V_{\mathrm{actual}}\) [mm/s] is defined by the following equation:

Here, the power transmission efficiency \(\eta\) [%] is defined as the ratio of the measured speed to the theoretical ideal propulsion speed (\(V_{\mathrm{ideal}} = 36.65\) mm/s), as expressed in the following equation:

This propulsion loss can be attributed to four primary factors. First, the power transmission efficiency was highly dependent on the tension of the gummy belt, and slippage occurred between the gears and the belt when the tension weakened. As shown in Table 2, the maximum fracture load of the candy axle is \(26.7{\pm}8.2\) N. The initial tension had to be minimized to prevent the fracture of the axles. This design constraint led to the inability to maintain a sufficient tension during driving, consequently causing the slippage. Second, due to the one-sided support of the motor-side gear, the design allowed the force to escape when a high load was applied, as the motor-side gear yielded to the reaction force from the axle gear. Third, when driving, the tension of the belt pulled the axles, generating a significant frictional resistance at the contact areas between the axles and the PLA body, which hindered the propulsion. Fourth, the inherent surface tackiness of the gummy material resulted in adhesion at the meshing points between the gears and the timing belt, which temporarily impeded a smooth rotational motion.

Thus, the elasticity of the gummy material induced minute deformations in the gear teeth and friction, resulting in a transmission loss of 14.1%. However, the primary motivation for selecting gummy as the material for the gears and belt in this robot was prioritizing its material toughness to prevent catastrophic failure of the entire system. If brittle materials such as cookies or candy were used, the high stress concentration at the gear meshing points would pose a significant risk of sudden tooth chipping or fracture. Therefore, this design adopted a gummy formulated with a high gelatin content to optimize its hardness as a gear while maintaining sufficient flexibility to withstand these stresses.

Edible bearings and edible lubricants should be introduced to address these issues and improve the transmission efficiency to the ideal value, besides reinforcing the support structure of the motor gear. Furthermore, applying a powder coating, such as potato starch, to the axle surfaces to reduce the friction coefficient can be extremely effective in improving the driving performance of future edible mobile robots. From the material design perspective, developing composite edible materials is a key future prospect. For instance, approaches such as adding dietary fibers to the gummy base to increase the elastic modulus while retaining its toughness, or employing a hybrid structural design—using hard candy as a rigid core and coating only the gear teeth surfaces with a thin gummy layer for shock absorption—could achieve an optimal balance between the transmission efficiency and durability.