Bipedal walking is the most automated and complex dynamic system among walking methods. Bipedal walking systems have very rich dynamic characteristics, have low requirements for the walking environment, can walk on flat ground, and can also walk on unstructured complex ground, and have good adaptability to the environment. Compared with other legged robots, bipedal robots have the characteristics of small support area, large change of support surface shape over time, and high relative position of center of mass. It is the most complex and most difficult dynamic system to control. However, due to the higher flexibility of bipedal robots than other legged robots, they have their own unique advantages and are more suitable for working with humans in human living or working environments without the need for large-scale modifications to these environments. For example, replacing workers in dangerous working environments (such as nuclear power plants), carrying goods on uneven ground, etc. In addition, future changes in the social environment make bipedal robots have great potential in nursing the elderly, rehabilitation medicine, and general household chores. Determination of the degrees of freedom of bipedal walking robots The mechanism of bipedal walking robots is the carrier of all components and is also the most basic and primary task in designing bipedal walking robots [1]. It must be able to perform basic functions such as moving forward, backward, left, and right, as well as climbing slopes and stairs. Therefore, the configuration of degrees of freedom must be reasonable. First, let's analyze the walking robot's motion process (forward) and walking steps: shifting the center of gravity to the right (first the right leg supports the body), lifting the left leg, lowering the left leg, shifting the center of gravity to the middle of both legs, shifting the center of gravity to the left, lifting the right leg, lowering the right leg, and shifting the center of gravity to the middle of both legs, a total of 8 stages. From the robot's walking process, we can see that: when the robot steps forward, the hip joint and ankle joint must each be equipped with a pitch degree of freedom to cooperate in realizing the movement of the supporting leg and upper body; to realize the transfer of the center of gravity, the yaw degree of freedom of the hip joint and ankle joint is essential; the robot sometimes has to turn to reach the target position, so a rotation degree of freedom on the hip joint is required. In addition, configuring a pitch degree of freedom at the knee joint can adjust the landing height of the swing leg, making it possible to go up and down stairs, and also realize different gaits. This ultimately determines that the hip joint is configured with 3 degrees of freedom, including roll, pitch and yaw, the knee joint is configured with one degree of pitch, and the ankle joint is configured with two degrees of pitch and yaw. Thus, each leg is configured with 6 degrees of freedom, and the two legs have a total of 12 degrees of freedom. The pitch degrees of freedom of the hip, knee and ankle joints work together to complete the robot's straight walking function in the longitudinal plane (forward direction); the roll degree of freedom of the hip joint can realize the robot's turning function; the yaw degrees of freedom of the hip and ankle joints work together to realize the center of gravity transfer function in the transverse plane. The robot's roll, pitch and yaw are defined as shown in Figure 1 [2]. [align=center] Figure 1 Schematic diagram of walking robot direction[/align] Wherein, the y-axis direction is defined as the forward direction, and the z-axis direction is defined as the robot's height direction. The overall configuration of the robot's degrees of freedom is shown in Figure 2. [align=center]Figure 2 Overall Configuration Diagram of Degrees of Freedom[/align] Power Source Selection Currently, there are many types of electric motors on the market that provide energy to robots: DC motors, AC motors, stepper motors, and servo motors. Because bipedal walking robots require high precision, and AC/DC motors turn on when powered and stop when powered off, it is difficult to perform position control on the robot; while stepper motors can work with a certain level of precision, they are open-loop systems and cannot meet the required precision. Therefore, this paper chooses to use servo motors. The relatively inexpensive servo motor used in this paper is the servo motor—the servo motor. Servos first appeared in model aircraft. In model aircraft, the flight attitude of the aircraft is achieved by adjusting the engine and various control surfaces. The working principle of an electric servo motor is shown in Figure 3. The servo motor controller generally uses PID control to meet the servo motor's static and dynamic performance requirements; the servo power amplifier generally consists of a pulse width modulator (PWM) and a switching control circuit; the DC servo motor is the actuator of the electric servo motor and can be a brushed or brushless DC motor; the reduction mechanism generally uses a worm gear or screw cylinder reduction mechanism. Since the robot produced in this study is a small bipedal walking robot with a very light weight and a desktop design for experimental use, the joints of the robot are driven by servo motors. [align=center] Figure 3 Block diagram of the working principle of electric servo motor[/align] Appendix Parameters of RC servo motor The characteristics of this type of motor are small size, light weight and simple control, and it is also relatively inexpensive. The appendix shows the parameters of the motor. The walking robot has 6 degrees of freedom for each leg. The high torque servo motor of HG14-M from Beijing Hanku Technology Co., Ltd. is used to drive each joint. Mechanism Design According to the requirements of this project, the mechanism of the robot is designed. Its main features are as follows: Structural symmetry is common in walking motion. Goldberg et al. [3] studied the symmetry in walking motion and found that there is a relationship between the symmetry of the body motion and the symmetry of the leg mechanism. In the single-leg support stage as shown in Figure 4, the symmetrical body motion requires the leg mechanism to be symmetrical as well; in the double-leg support stage as shown in Figure 5, the symmetrical body motion does not necessarily require the symmetry of the leg mechanism unless there are additional constraints. Based on this, the author also adopted a symmetrical arrangement in the structural design [4]. The frame design effectively utilizes the size of the RC servo motor and makes the range of motion of the motor as consistent as possible with the range of motion of each joint. A multi-joint structure is adopted. The walking mechanism can realize functions such as moving forward and backward on flat ground, moving sideways on flat ground, turning, going up and down stairs, and climbing slopes. The entire structure uses 1mm aluminum alloy (LY12) sheet metal material. This material is lightweight, has high hardness, and although its strength is not as good as steel, it is much higher than ordinary aluminum alloy. Moreover, this material has the characteristics of high elastic modulus and high density. Since the joints of the robot are driven by RC servo motors, in order to reduce the size and weight of the robot, the structure of the robot is made into a frame type. The frame design effectively utilizes the size of the RC servo motor and makes the range of motion of the motor as consistent as possible with the range of motion of each joint. Control system Scheme conception Since the robot mechanism uses 12 servo motors, the control system is to realize the function of driving these 12 servo motors at the same time. As described above, the control signal of the servo motor is a pulse width modulation (PWM) signal with a period of 20ms, in which the pulse width ranges from 0.5ms to 2.5ms, and the corresponding position of the servo disk is 0 to 180°, which changes linearly [5]. That is to say, if a certain pulse width is provided, its output shaft will remain at a corresponding angle, no matter how the external torque changes, until a pulse signal of a different width is provided, it will change the output angle to a new corresponding position. [align=center] Figure 4 Single-leg model[/align] [align=center] Figure 5 Bipedal walking robot dual-leg model[/align] The traditional method of generating PWM waves is achieved by a large number of discrete components, and the generated pulse frequency and width are often not very accurate, making it difficult to achieve precise control of the servo motor. At present, there are many methods for generating PWM waves: the most direct method is to use the PWM port of the microcontroller itself to generate waveforms, but this method is limited by the internal resources of the MCU and can only achieve the output of 2 to 4 PWM waves, which is obviously not enough for situations that require multiple servos. Another method is to use the idea of ​​time-division multiplexing to generate 7 channels of PWM waves for controlling Futaba servos using a single interrupt of the microcontroller. Although this method achieves the control of 7 servos, it can only achieve the control of 7 servos and is only for the control of specific servos. The control accuracy is not high, and its application in some important situations is limited. Another method is to use the pure software loop counting method of the microcontroller or the combination of hardware timing and software counting. Without adding any hardware interface, multiple PWM waves can be output [6]. However, this method occupies a lot of MCU operation time and basically cannot handle other tasks, and the accuracy is not high. In addition, some digital signal processing chips have integrated PWM waveform generation function on the chip. Only the register parameters need to be set to obtain PWM output. However, in some situations where only simple motor control is needed, the relatively expensive digital signal processing chip is not required from the perspective of cost. This paper uses a 51 microcontroller and a complex programmable logic device (CPLD) to realize the generation of PWM. Because CPLD has its own parallel processing capability and a large number of IO interfaces, it can control dozens or even hundreds of servo motors to work at the same time, which can leave some space for subsequent work. However, since CPLD does not have transaction processing capability, MCU is still needed to work together in actual applications. This paper uses a 51 series microcontroller and CPLD to control the servo motors. In addition, the use of microcontroller can also leave space for subsequent sensor feedback processing. [align=center] Figure 6 Control system structure diagram[/align] As shown in Figure 6, the control system solution of "upper computer + serial port + lower computer" is selected. The main function of the upper computer control software is to plan and interpolate the predetermined robot action, and then send it to the lower computer according to a certain time interval and sequence to realize the robot joint position and approximate speed control; the main function of the lower computer is to receive the position signal sent by the upper computer, generate PWM wave according to the signal requirements, control the movement of each joint servo motor of the robot, and make the robot complete the skating action according to the action plan. Correspondingly, the lower computer is mainly composed of serial communication, data scheduling and 12 servo motor drive modules[7]. [align=center]Figure 7 Schematic diagram of a 12-channel servo drive in a CPLD[/align] Figure 7 shows the schematic diagram of a 12-channel servo drive in a CPLD. The CPLD communicates with a 51 microcontroller through a simple interface, storing the PWM signal data of the 12 servos to be driven in the data storage area. The digital PWM generator then drives the 12 servos to the required angle. When a new angle is needed, new PWM signal data is transmitted from the 51 microcontroller to the data storage area for updating, thus updating the digital PWM generator and driving the servos to a new angle. Hardware Design Power Supply: To avoid voltage fluctuations from the servo power supply interfering with the control circuit, the power supplies for the control circuit and servos are isolated, i.e., powered separately. The control circuit power supply uses a 9V output AC-DC transformer, which provides 5V power after passing through a 7805 chip. The servo power supply provides an interface for connecting an external 6V DC power supply. The control chip module includes a microcontroller, clock circuit, reset circuit, external program memory chip expansion, and a large-scale CPLD chip. The microcontroller used is the Atmel AT89S52, an 8-bit high-performance embedded controller with integrated 8KB of in-system programmable flash memory, 256 bytes of RAM (64 bytes addressable), 32 I/O ports, 3 programmable timers, 8 interrupt sources, 6 interrupt vectors, and 1 watchdog timer. The clock circuit provides the system time base, using an 11.05296MHz crystal oscillator. Additionally, an 8KB × 8-bit external memory chip, the 2864, is also included. The CPLD chip used is the Altera EPM7128. The serial communication module is mainly used for serial communication between the AT89S52 microcontroller and the PC. Since the PC's COM port conforms to the RS-232 standard, while the serial interface on the AT89S52 microcontroller is at CMOS level, level conversion is required when communicating between RS-232 and CMOS levels. Therefore, the MAX232 chip is used for level conversion in the design. The servo motor drive control signals for the 12 servo motors come from the I/O ports of the CPLD chip (pins 30, 31, 33, 34, 35, 36, 37, 39, 40, 41, 44, 45). To prevent interference, the 13 servo motor control signals and drive circuits should be opto-isolated by a TLP-521. The isolated control signals must also be connected to an LM324 comparator to eliminate glitches, increase signal stability, and improve the output current, ensuring the servo motors operate correctly without unnecessary jitter. Figure 8 shows the hardware circuit board design diagram of the control system. [align=center]Figure 8. Robot Controller Circuit Diagram[/align] Conclusion This paper discusses the installation method of servo motors, the design of the frame, and the fabrication of an economical bipedal walking robot capable of motion control via servo motors. Furthermore, it realizes the concept of controlling 12 RC servo motors using a microcontroller and CPLD. Future research will explore the application of inverse kinematics principles, pre-determining the motion trajectories of various robot parts, calculating the rotation angles of each joint, and then using a control system to obtain a control algorithm to realize the robot's actual walking process.