In this paper, the authors present the process of modeling, building, and testing two prototypes of tetrahedral robots with omnidirectional locomotion units. The paper begins with a detailed description of the first tetrahedral robot prototype, highlighting its strengths as well as the limitations that led to the need for improvements. The robot’s omnidirectional movement allowed it to move in all directions, but certain challenges related to stability and adaptability were identified. The second prototype is presented as an advanced and improved version of the first model, integrating significant modifications in both the structural design and the robot’s functionality. The authors emphasize how these optimizations were achieved, detailing the solutions adopted and their impact on the robot’s overall performance. This paper includes an in-depth comparative analysis between the two prototypes. The analysis highlights the considerable advantages of the second prototype, demonstrating its superiority. The conclusions of the paper summarize the main findings of the research and emphasize the significant progress made from the first to the second prototype. Finally, future research directions are discussed, which include refining control algorithms, miniaturizing the robot, improving structural performance by integrating shock-absorbing dampers, and integrating lighting systems and video cameras. The field of mobile robotics is a vast and dynamic domain that is focused on developing robots that are designed to perform tasks that either ease the workload or eliminate risks to the human operator’s life. Their utility is demonstrated in a variety of applications, such as search and rescue operations, interventions in hazardous environments, medical assistance, and more [ 1, 2, 3]. Tetrahedral-structured robots also fall within the spectrum of mobile robotics. These robots take the shape of a polyhedron with four triangular faces, with their main advantage being stability and flexibility in movement. The concept of the tetrahedral robot attracts attention due to its adaptability in various fields, from inspections in flat environments, where access is difficult or dangerous, to rugged terrains. Unlike other categories of mobile robots, it is important to mention that the most significant feature of tetrahedral robots is their symmetrical geometric shape, which allows the robot to move regardless of its position, eliminating the issue of immobility in the event of a rollover. Tetrahedral robots with omnidirectional locomotion units are designed specifically for applications in areas where the risk of rolling exists. Due to their symmetrical structure, they achieve a stable position even in the event of falls and rolls. The robots proposed by the authors will be able to perform inspections/surveillance in both enclosed and open spaces and will transmit information to the human operator. For example, surveillance inside museums, building inspections, monitoring spaces with potential degradation risks, etc. The literature mentions several achievements in the field of tetrahedral robots. The concept of a tetrahedral robot with various types of locomotion units attached is presented [ 4], as well as the development and testing of a tetrahedral stabilopod robot prototype with Omni-Ball locomotion units [ 5], and its reconfiguration from a tetrahedral structure to a flat structure to enable movement in narrow spaces [ 6]. In the realm of tetrahedral robots, there are also modular robots, such as the Odin robot [ 7], which can reconfigure by combining multiple tetrahedral structures [ 8, 9], or the modular ARTS robot, which is based on adaptive tetrahedral elements [ 10]. Another example presented in the specialized literature that demonstrates the feasibility of these types of structures is the construction of the Tetrahedron Walker Robot, which has the ability to perform autonomous movement [ 11]. Among the latest publications regarding tetrahedral robots is the soft-legged robot, which has the ability to perform omnidirectional movements by bending and is powered pneumatically [ 12]. To maximize the efficiency of this type of robot, various types of locomotion units have been integrated, which have the capability to change direction without the need for additional maneuvers. These locomotion units are referred to in the specialized literature as standard wheels or omnidirectional wheels and can be differentiated by several characteristics, such as shape, design, complexity, etc. The design of these omnidirectional wheels consists of a wheel surrounded by rollers. The wheel performs the active movement while the rollers on the circumference of the wheels execute the passive movement. In the specialized literature, several types of omnidirectional locomotion units can be identified, such as the continuous alternative wheel [ 13, 14], which focuses on reducing the gaps caused by the spaces between the rollers; the double or single universal wheel [ 15, 16], with rollers arranged circularly on the wheel’s surface; the large number of passive rollers designed by Laquos [ 15] and Wesn [ 17] that attempt to reduce vibration issues by minimizing the space between the rollers; the omnidirectional wheel with a spherical geometry of the Omni-Ball type [ 15], which offers smooth movement with greater freedom of motion over a wide variety of terrains; and the Mecanum wheel with rollers arranged at a 45° angle [ 18, 19, 20]. This paper is structured as follows: The first part describes the development of the first prototype, starting from the initial phase of 3D modeling and concluding with the testing of the completed prototype. The second part refers to the development of the second prototype, following the same stages. The third part describes the kinematic analysis of the robots with the three omnidirectional locomotion units located at the base of the robot and in contact with the ground surface. The final part of the paper analyzes the two prototypes developed, emphasizing the improvements made to the second prototype compared to the first, followed by the general conclusions of the paper and future research directions. To establish a valid option for practical implementation, two initial designs of tetrahedral robots were modeled. These two designs have a tetrahedral structure in two main geometric forms—the pyramidal shape and the stabilopod shape. The first design, shown in Figure 1a, consists of a closed pyramidal casing on three of the four faces, with a detachable face that allows access to the interior of the robot. The corners of the pyramid are beveled, where the locomotion units are mounted. The second design, shown in Figure 1b, is made from a small trapezoidal body to which four pipes are attached as legs, with the locomotion units mounted at the ends of these pipes. The locomotion unit modeled for both designs is a complex one, designed based on the double universal wheel model but adapted to a spherical geometry of the Omni-Ball type. It consists of four elements that form two identical hemispheres, which are joined together with rollers across the entire surface of the hemispheres; these serve both to facilitate passive movement and to provide an esthetic function. In the components of the locomotion units, there is a black rubber band mounted over the rollers, as shown in Figure 2. This serves as a tread, and the material from which it is made eliminates the issue of roller slippage on the contact surface. The second design of the tetrahedral robot focused on defining the pyramidal geometric structure, consisting of four triangular faces joined together and beveled at the corners, creating an optimal surface for attaching the locomotion units. In the center of the robot’s body, a horizontal plate was placed for securing the electronic components. In Figure 3, the final design of the robot can be observed. This represents the design that will subsequently be developed and tested. The locomotion units ( Figure 4a,b) used for the final design of the tetrahedral robot show slight modifications compared to the first modeled variant in Figure 1. They are constructed by connecting two symmetrical wheels. Each wheel consists of a set of 16 rollers mounted on the circumference of the wheel, over which a rubber band is attached to reduce friction between the ground and the locomotion unit. Additionally, this band allows for the passive movement of the locomotion unit in a direction perpendicular to the active movement of the locomotion unit. The connection of the locomotion units to the motor shaft is made through a shaft that passes through the center of the locomotion units. The final design of the tetrahedral robot presented in Figure 3 highlights both the simple, accessible, and easy-to-construct design of the robot’s body, as well as the rigidity, stability, and complexity of the locomotion units. The body of the prototype was made from PMMA, a material chosen for its accessibility and available processing technology. The laser cutting process was selected due to the high precision it offers. The locomotion units were made from polylactic acid (PLA) using 3D printing technology, which allows for the production of precise components. This process is essential for creating parts with complex geometry and ensures efficient and rapid execution in terms of resources and costs. Figure 5a presents the construction of the first tetrahedral robot prototype with omnidirectional locomotion units, followed by its testing and analysis, while Figure 5b presents the controller that allows for the remote operation of the robot. To implement the control of the robot, Arduino software, version number 1.8.19 was used. The first step in achieving the robot’s control is to create a block diagram, as shown in Figure 6, as well as the electrical diagram ( Figure 7), which provides an overview of the connections between the component elements. The diagram is divided into two parts—the left side refers to the component elements inside the robot, which are used for controlling the motors, while the right side refers to the robot controller operated remotely with the help of a joystick. The robot uses four DC motors. To control these from within the robot, two L298N dual motor drivers (STMicroelectronics, Geneva, Switzerland) and an Arduino UNO development board (Arduino SA, Ivrea, Italy) were required, both connected to a 9V battery. The system communicates with the controller via a wireless module (Nordic Semiconductor ASA, Trondheim, Norway). The controller is built using an Arduino Nano development board (Arduino SA, Ivrea, Italy) powered by a 5V battery, and the robot’s movements are executed using a joystick (Keyes, Shenzhen, China). The testing of the robot was conducted in three stages, which refer to the prototype’s forward movement, lateral movement, and movement on a ramp with a 15° incline, in three positions, as shown in Figure 8. Table 1 presents the time in which the robot reaches maximum efficiency based on three types of movement—forward, lateral, and climbing the ramp. For forward and lateral movement, the distance covered by the robot is 1 m, while for ramp climbing, the distance is 40 cm. These data indicate the robot’s performance in terms of speed and efficiency for each of the three cases. It can be observed that the robot covered the distance of 1 m most quickly in the lateral direction, while the movement on the ramp took the longest time. Following testing, the main deficiencies of the constructed prototype were identified. These refer to the robot’s rigidity, sealing issues, instability, and insufficient contact between the locomotion units and the movement surface. The insufficient rigidity of the robot is caused by the material used for constructing the robot’s body, as well as the joints between the panels, which lead to the formation of gaps at the connections. Additionally, the instability of the center of gravity is determined by the positioning of the horizontal plate inside the robot, which becomes problematic during positional changes. Furthermore, insufficient contact between the locomotion unit and the ground reduces the efficiency of movement, as some of the locomotion units remain unused. Following the completion of the first prototype, it was concluded that major improvements need to be made regarding the assembly of the robot’s body components, focusing the electronic components into a core so that the robot’s center of gravity is positioned at the center of the structure, as well as selecting materials and reinforcing the structure. In this regard, three designs were modeled in 3D, as presented in Table 2, which were also analyzed. It can be observed that in addition to the advantages presented in Table 2, the designs have certain major disadvantages that lead to the establishment of an optimal final model. The first design shows deficiencies in terms of stability during rolling because the central core is not symmetric, which leads to the robot’s imbalance. Another important aspect of this model is the absence of mounting points for the motors, which renders the robot non-functional. The third disadvantage relates to the holes distributed across the entire surface of the skeleton, which, although intended to reduce weight, decrease the robot’s strength. The second design exhibits the smallest number of weaknesses among all three designs and is the one that will be practically realized. The main disadvantage of this model is its high weight compared to the other two models, as well as the increased attention required for the distribution of the components within the body to avoid the robot’s imbalance. This structure (modeled in Figure 9) presents multiple advantages, including its stability and rigidity, which provide a solid operational base in the form of the fixed attachment of the motors to the robot’s body, ensuring the precise control of the locomotion units, as well as the durability and strength of the materials, contributing to its reliable operation over an extended period. The structure of the second prototype consists of two subassemblies, which are shown in Figure 10a,b. The first modeled subassembly consists of six rectangular components (which form the skeleton of the structure) and four disks that are used as fastening elements for the six edges. The second subassembly represents the core of the robot’s body, consisting of four triangular plates (with beveled tips) and four fastening disks, inside of which are the electronic components that are evenly distributed to maintain a constant center of gravity. The connection between the two subassemblies is made through the motors as well as the hub of the locomotion units, creating a rigid structure. The third design (from Table 2) has a distinct geometric structure compared to the other two models, namely the stabilopod shape. Although this model may be one of the optimal designs, it has a smaller weight and smaller dimensions compared to the other two models presented. The dampers around the motors reduce the vibrations caused during the impact between the robot and the contact surface, but the costs are significantly higher. In addition to the three 3D-modeled body designs of the robot, two omnidirectional mobility units were also modeled—one of the Wire Sphere type ( Figure 10) and one of the Omni-Ball type [ 4]; these were also manufactured. The two designs presented in Figure 11 and Figure 12, although they have a spherical shape, possess several characteristics that differentiate them, such as their structure, material, and applicability. The Wire Sphere-type locomotion unit ( Figure 11) is made of wires mounted on a hub, forming an open structure. It is constructed from lightweight yet durable metals, making it especially suitable for rough environments. Active displacement is achieved through a hub connected to a motor, which drives the locomotion unit. Passive displacement occurs by sliding over the surface of the terrain, without the direct action of the motor. This type of locomotion allows the unit to adapt to various types of terrain. One of the main advantages of these locomotion units lies in their wire structure, which acts as a damping system during collisions. The Omni-Ball locomotion unit ( Figure 12) is made up of two identical hemispheres, each mounted on a bearing ( Figure 12b) that contributes to passive movement. They are subsequently mounted on a shaft that connects the locomotion unit to the motor, making the structure of the locomotion unit a closed one. The Omni-Ball is made from polylactic acid (PLA), which is known for its strength and low environmental impact. Due to its design, the Omni-Ball allows for efficient and controlled maneuverability, making it ideal for applications that require precise movements in space-constrained environments. The hub of the locomotion unit is an essential component in its construction, as it serves as the support on which the two hemispheres that make up the locomotion unit are mounted. The locomotion unit performs active movement through the motors. Passive movement is facilitated by the two hemispheres that make up the locomotion unit in a direction perpendicular to the drive motor’s hub. After comparing the three designs of the robot’s body (from Table 2) and the two locomotion units, it was determined that the best option in terms of stability, rigidity, ease of construction, and reduced costs is the second design of the robot combined with the OmniBall-type locomotion units. The manufactured prototype is presented in Figure 13. An essential part of the practical implementation refers to the manufacturing of components; the 3D printing process is illustrated in Figure 14. All of these were accomplished using a Bambu Lab A1 3D printer (Bambu Lab, Shenzhen, China), utilizing a durable material, namely polylactic acid (PLA), to enhance the robot’s strength. Another important aspect of the practical implementation refers to both the assembly of the mechanical components and the installation of the electronic components according to the diagrams presented in Figure 15. The proposed block diagram aims to provide a simplified overview of the robot’s structure and functionality, as well as the electrical schematic that accurately presents the connections between these elements. The schematics include all electronic components and their connections, serving as a fundamental tool for enabling the robot’s locomotion. The robot is controlled by the main development board, an Arduino Nano (Arduino SA, Ivrea, Italy), to which two dual drivers are connected for controlling the four motors. The motors are powered by a 9V battery, while the development board is powered by a 5V battery, as shown in Figure 15a,b. The robot’s movements will be controlled using a joystick, and the MPU 6050 orientation sensor (InvenSense (now part of TDK Corporation), San Jose, CA, USA) will enable the robot to operate regardless of its position. The final step in the practical implementation focuses on testing and analyzing the robot’s functionality. The testing of the second prototype was conducted in five stages ( Figure 16). For this purpose, the robot was required to move from an initial position to a final position across all four sides of its structure, observing the correct operation of the locomotion units as the robot changed positions. The four locomotion units were labeled A, B, C, and D. In Figure 16a, the robot begins its movement on the locomotion units ABC located at the base of the robot, followed by a return to move on the locomotion units ABD ( Figure 16b). This is then followed by a return on BCD ( Figure 16c) and then on ACD ( Figure 16d). Finally, in Figure 16e, the robot is repositioned to the same position as the initial one, specifically having the locomotion units ABC at its base. The correct movement of the locomotion units, based on their position, is attributed to the MPU 6050 orientation sensor [ 24], which is designed to detect the robot’s orientation in space. By utilizing this sensor, only the locomotion units at the base of the robot are activated. As mentioned, the tetrahedral robot with omnidirectional locomotion units is characterized by its symmetrical structure, allowing movement in any direction without the need for additional maneuvers. Considering that the robot has a pyramidal shape, the kinematic analysis will be conducted for the situation in which the robot is in contact with the locomotion surface using the three locomotion units positioned at its base. In Figure 17 the kinematic diagram of the robot moving on a flat surface is presented. Three locomotion units are depicted at the extremity of one face of the robot. In the kinematic modeling, a relationship will be established on the flat surface between the velocities of the locomotion units φ˙1, φ˙2, φ˙3" role="presentation"> φ˙1, φ˙2, φ˙3 φ ˙ 1 ,   φ ˙ 2 ,   φ ˙ 3 and the velocity of the robot x˙, y˙, θ˙" role="presentation"> x˙, y˙, θ˙ x ˙ ,   y ˙ ,   θ ˙ . In this regard, we define the base coordinate system as Oxy in the robot’s surrounding environment and the system attached to the robot as OLxLyL [ 21]. The locomotion units are arranged at 120° angles, positioned by α 1, α 2, and α 3. Measuring these angles in the trigonometric sense from x L, the angles are α1=0°, α1=120°, α1=240°" role="presentation"> α1=0∘, α1=120∘, α1=240∘ α 1 = 0 ° ,   α 1 = 120 ° ,   α 1 = 240 ° [ 21]. The angular velocities of the robot’s wheels are determined using the following relationship [ 21]: φ˙1φ˙2φ˙3=1r·−sin⁡θcos⁡θR−sin⁡(θ+α2)cos⁡(θ+α2)R−sin⁡(θ+α3)cos⁡(θ+α3)Rx˙y˙θ˙" role="presentation"> ⎡⎣⎢⎢⎢⎢φ˙1φ˙2φ˙3⎤⎦⎥⎥⎥⎥=1r·⎡⎣⎢⎢⎢−sinθ−sin(θ+α2)−sin(θ+α3)cosθcos(θ+α2)cos(θ+α3)RRR⎤⎦⎥⎥⎥⎡⎣⎢⎢⎢x˙y˙θ˙⎤⎦⎥⎥⎥ φ ˙ 1 φ ˙ 2 φ ˙ 3 = 1 r · − s i n ⁡ θ cos ⁡ θ R − sin ⁡ ( θ + α 2 ) cos ⁡ ( θ + α 2 ) R − sin ⁡ ( θ + α 3 ) c o s ⁡ ( θ + α 3 ) R x ˙ y ˙ θ ˙ (1) where—R is the distance from the center of the robot to the center of the locomotion units, and r is the radius of the locomotion units. Using the notations φ˙1=ω1, φ˙2=ω2, φ˙3=ω3" role="presentation" style="position: relative;"> φ˙1=ω1, φ˙2=ω2, φ˙3=ω3 φ ˙ 1 = ω 1 ,   φ ˙ 2 = ω 2 ,   φ ˙ 3 = ω 3 , the following is obtained: ω1ω2ω3=1r·−sin⁡θcos⁡θR−sin⁡(θ+α2)cos⁡(θ+α2)R−sin⁡(θ+α3)cos⁡(θ+α3)Rx˙y˙θ˙" role="presentation" style="text-align: center; position: relative;"> ⎡⎣⎢⎢⎢ω1ω2ω3⎤⎦⎥⎥⎥=1r·⎡⎣⎢⎢⎢−sinθ−sin(θ+α2)−sin(θ+α3)cosθcos(θ+α2)cos(θ+α3)RRR⎤⎦⎥⎥⎥⎡⎣⎢⎢⎢x˙y˙θ˙⎤⎦⎥⎥⎥ ω 1 ω 2 ω 3 = 1 r · − s i n ⁡ θ cos ⁡ θ R − sin ⁡ ( θ + α 2 ) cos ⁡ ( θ + α 2 ) R − sin ⁡ ( θ + α 3 ) c o s ⁡ ( θ + α 3 ) R x ˙ y ˙ θ ˙ (2) The first difference between the two robot prototypes is their design. The first design is much simpler, consisting of four identical plates connected and a supporting plate inside the robot intended for securing the electronic components. The second design includes two subassemblies—the core and the skeleton of the robot—resulting in a more robust body [ 21]. Another difference lies in the rigid structure due to the mechanical fastenings and the materials used to construct the robot’s body. The transition from manufacturing the robot using PMMA to PMA creates a structure that is more rigid and resistant. Compared to PMMA, PMA is much more durable and resistant to rigid mechanical clamping conditions. While PMMA can break or crack when subjected to mechanical stress or rigid clamping, PMA enables durable mechanical clamping without the risk of cracking, as a result of its greater rigidity and improved structure. Therefore, PMA is much better suited for applications requiring mechanical strength and long-term durability. The uniform distribution of electronic components is possible in the design, which allows for their fixation on each plate of the core, ensuring a stable center of gravity. In contrast, in the first design, where the components are mounted on a single horizontal plate, the rolling of the robot causes the destabilization of the center of gravity. Another major difference in the comparison between the two prototypes is the addition of the MPU 6050 orientation sensor in the second prototype, which enables the correct functioning of the mobile tetrahedral robot [ 21]. The first prototype cannot move in all four scenarios illustrated in Figure 3. It can be observed that another difference between the two robot prototypes is the structure of the locomotion units. Considering the two locomotion units developed, the first unit has a complex structure, and due to the discontinuous contact during movement (caused by the gaps between the rollers), it may introduce vibrations. This locomotion unit consists of two identical parts connected to each other. Each part is designed based on the model of a double omnidirectional wheel. Because the mobile robot has a tetrahedral structure, and the locomotion units are positioned at an angle, it does not maintain full contact with the ground. This causes one of the two parts that make up the unit to be ineffective on a flat surface. Regarding the Omni-Ball units, one specific design characteristic is their spherical shape. This contributes to improving contact with the locomotion surface by ensuring that they maintain constant contact with the surface, regardless of the rotation position. Thus, this unit provides a stable and continuous contact surface, which enhances the robot’s motion control. The comparison of the advantages of the two prototypes is presented in Table 3. This table highlights the improvements introduced by the second design. To correctly analyze the two prototypes, the authors use metrics that allow for an objective comparison [ 25, 26, 27]. In this context, we use the existing information to evaluate the metrics associated with the two tetrahedral robot prototypes with omnidirectional locomotion units. The first relevant metric in comparing the two robots is structural stability. In this regard, the symmetrical distribution of weight is a crucial factor for the tetrahedral structure. The electronic components of the first prototype are concentrated on a central board located at the center of the robot’s body, which may cause instability, while the second prototype has the electronic components distributed across all faces of the robot’s body, helping to maintain its stability. The second relevant metric for the two prototypes concerns energy consumption. The addition of the MPU 6050 orientation sensor in the second prototype contributes to the efficient operation of the locomotion units. The first prototype, lacking this sensor, has deficiencies in this regard, as all four locomotion units at the pyramid’s tips are active simultaneously during movement, resulting in a higher energy consumption. In contrast, after integrating the sensor into the second prototype, the robot gains the ability to detect orientation, so only the three locomotion units at the base of the robot are activated, automatically reducing energy consumption. The last metric used in the comparison of these robots is maneuverability. The Omni-Ball locomotion unit used in the second prototype represents a superior solution in terms of maneuverability, featuring a design that allows the robot to move easily in any direction without interrupting contact with the locomotion surface. The double universal locomotion unit used in the first prototype is based on rollers and may cause vibrations during operation due to the gaps between the rollers. The objective of this paper is the development of two tetrahedral robot prototypes with omnidirectional locomotion units. This paper describes the two prototypes developed from the initial phase to their practical implementation and testing. Following the analysis and testing of the first design of the tetrahedral robot, it was concluded that a second design was necessary to which substantial improvements could be made regarding design and functionality. The new structure contributes to its superior performance and increased adaptability in various applications and operational environments. Both models have been practically implemented, providing a solid foundation for future research in the field of mobile robotics. The main improvements pertain to the design, strength, complexity, and functionality of the robot. The second prototype includes the MPU6050 sensor. The implementation of this MPU6050 orientation sensor is one of the most important features of the robot, due to its advanced capabilities in motion and orientation detection. Three metrics were added to analyze the differences between the two prototypes. The miniaturization of the robot is an essential process for accessibility in confined spaces. This transformation involves not only reducing the overall dimensions of the robot, but also adjusting various components to maintain functionality and operational efficiency. Reducing the structural dimensions of the robot is a major challenge that can be addressed by using new materials, such as carbon fiber or lightweight metal alloys. These materials reduce the weight while ensuring increased strength for the robot’s structure. Another challenge involves the miniaturization of electronic components. This can be achieved by selecting smaller microcontrollers or by consolidating the circuitry into a single compact unit. Regarding power supply, the solution is to use lithium-polymer batteries, which are characterized by their high energy density while having a reduced size and weight, making them ideal for this type of application. Additionally, solutions will be identified to enhance mobility. These solutions could include optimizing the structural design and using lighter or more flexible materials. The integration of lighting systems and video cameras is an important factor, considering the environments in which the robot is applicable, such as dark or complex environments. The addition of lighting systems and video cameras can affect the weight distribution and increase the robot’s power requirements, necessitating adjustments to maintain balance and energy efficiency. Since the balance of a tetrahedral robot is crucial for stability and efficient movement, careful placement of the new components within the robot is proposed to prevent imbalance and to achieve the optimal counterbalancing of the entire system. Regarding power requirements, the new lighting equipment and video cameras impose additional energy consumption. The proposed solutions include using a higher-capacity battery that is capable of supporting the entire system, thus optimizing energy reserves, and implementing the intelligent management of the lighting and video systems, activating them only when necessary. Attaching dampers to enhance durability and shock absorption during movement is another important aspect for ensuring a long lifespan for the robot. Writing—original draft, A.-C.S. and M.O.T. All authors have read and agreed to the published version of the manuscript. This research received no external funding. Data are contained within the article. The authors declare no conflict of interest. Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. Abstract In this paper, the authors present the process of modeling, building, and testing two prototypes of tetrahedral robots with omnidirectional locomotion units. The paper begins with a detailed description of the first tetrahedral robot prototype, highlighting its strengths as well as the limitations that led to the need for improvements. The robot’s omnidirectional movement allowed it to move in all directions, but certain challenges related to stability and adaptability were identified. The second prototype is presented as an advanced and improved version of the first model, integrating significant modifications in both the structural design and the robot’s functionality. The authors emphasize how these optimizations were achieved, detailing the solutions adopted and their impact on the robot’s overall performance. This paper includes an in-depth comparative analysis between the two prototypes. The analysis highlights the considerable advantages of the second prototype, demonstrating its superiority. The conclusions of the paper summarize the main findings of the research and emphasize the significant progress made from the first to the second prototype. Finally, future research directions are discussed, which include refining control algorithms, miniaturizing the robot, improving structural performance by integrating shock-absorbing dampers, and integrating lighting systems and video cameras. Keywords: mobile; robot; tetrahedral shape; omnidirectional wheel; omnidirectional movement