Design of humanoid robot control system based on PC/104 and single chip microcomputer

0 Introduction

As a complex system that crosses various disciplines, more and more researchers use robots as experimental platforms because they integrate knowledge from many aspects, including mechanical structure design, control system construction, information collection and processing, kinematics and dynamics analysis, and artificial intelligence. Humanoid robots have gradually transformed from simply simulating human appearance, movements, and walking to human thinking, vision, touch, and intelligence. This has put forward higher requirements for the entire robot system, not only to further improve the mechanical structure and installation of the robot, but also to enhance the function and processing capability of the control system.

As for the control system, the control chips commonly used in humanoid robots are DSP, ARM or some other single-chip microcomputers. In order to further enhance the scalability of the robot, the embedded system PC/104 is used as the main control computer of the robot. It has the advantages of good real-time performance, low cost and miniaturization. It overcomes the shortcomings of insufficient functions of traditional single-chip microcomputer-based control systems and non-real-time characteristics of PC-based control systems, and has broad prospects in the application of humanoid robots.

1 Humanoid robot structure and control system

The robot has 21 degrees of freedom, including 2 degrees of freedom in the head, which can realize the pitch and left and right deflection of the head. A CCD camera is installed on the head, and it is equipped with a visual acquisition card and a visual processing computer, which can realize the recognition and positioning of the target and directly provide target information to the main control computer. Each arm has 3 degrees of freedom, which can complete actions such as extension and bending. It can provide support after the robot falls, allowing the robot to stand up by itself. There is 1 degree of freedom on the waist, which realizes the forward and backward tilt of the humanoid robot's trunk, which is convenient for the robot to adjust the center of gravity when walking or performing hand actions, and enhances the controllability and stability of the robot. There are 6 degrees of freedom in the lower limbs, including 2 degrees of freedom at the ankle joint and 3 degrees of freedom at the hip joint. It is similar to the structure of human legs and can flexibly complete various movements of the lower limbs. The entire structure of the humanoid robot adopts a frame structure, which is conducive to reducing the weight of the robot structure, improving the robot's carrying capacity, and providing more space for the improvement of the robot control system. As shown in Figure 1, a real picture of the humanoid robot.

The humanoid robot control system uses the ACS-4051VE PC/104 mainboard module as the main controller, directly connects the camera via USB, and connects the joint controller via an RS 232 serial port to realize the communication between the main control computer and the joint controller. The drive module and the joint controller are integrated on a PC board, mainly realizing the generation of PWM waves and driving the motor to rotate. The ACS-4051VE mainboard integrates the Intel 82559ERl0/100 Mb/s Ethernet card, and an external wireless network card can realize communication with an external wireless network. The humanoid robot control system is generally divided into two parts: the main controller module and the joint controller module. Its overall structure is shown in Figure 2.

Key features include:

(1) High-speed, pipelined 8051-compatible CIP-51 core (up to 25 MIPS), 70% of the instructions execute in 1 or 2 system clock cycles, which can meet the needs of joint controllers.

(2) There are 4 general 16-bit counter/timers, 16-bit programmable counter/timer array, 5 capture/compare modules, and 29 port I/O. By programming the chip and reasonably allocating the comparator and I/O port, 21 PWM waves can be generated on the C8051F310 chip. Since the microcontroller outputs the control quantity in digital form, it must be converted into analog control quantity through D/A conversion and driven by the servo amplifier.

Here, MAXIIM's 12-bit serial D/A chip MAX531 is used as the digital/analog conversion chip. The MAX531 works in bipolar voltage mode, and its output analog quantity ranges from -2.048 to +2.048 V with an accuracy of 1 mV. The output analog quantity is amplified by the operational amplifier and enters the servo amplifier to drive the motor.

C8051F310 is the control core of the joint controller. It is mainly responsible for the generation of 21 PWM channels. Four general 16-bit counters/timers and five capture/compare modules are integrated in the C8051F310 chip. One counter/timer and one compare module are used to control 6 I/O ports. The other three counters/timers and three comparators control 15 I/O ports to achieve the generation of 21 PWM waves. Here, the generation of 6 PWM waves is used to illustrate the use of C8051F310 to implement the circuit. The circuit diagram is shown in Figure 4: The CEXn pin generates pulse width modulation PWM output. The frequency of the PWM output depends on the time base of the PCA counter/timer. The module's capture/compare register PCA0CPLn is used to change the duty cycle of the PWM output signal. When the joint controller receives the given 6 motor rotation angle sequence data. The software arranges the 6 data from small to large, and calculates the difference between two adjacent numbers in turn, and arranges them according to the smallest number, the difference between the first two numbers to the difference between the last two numbers, and maps the addresses of the cross switches corresponding to the data from small to large in turn.

The program puts the first minimum angle number into the lower 8 bits of the comparison register PCA0CPLn. When the low byte of the PCA counter/timer (PCAOL) is equal to the value in PCA0CPLn, the output on the CEXn pin is set to "1". At the same time, the program puts the second data, i.e. the difference, into the comparison register PCA0CPLn, clears the PCA counter/timer, and sets the cross switch to the corresponding output pin. When the low byte of the PCA counter/timer (PCAOL) is equal to the value in PCA0CPLn again, the output on the CEXn pin is set to "1", until this set of data is completed. The count value in PCAOL overflows, the CEXn output is reset, and the second round of PWM wave generation is prepared.

3 Experiments

3.1 Image acquisition and processing

In order to enable the robot to achieve the predetermined goal, the software system must be designed and planned. The main control computer is installed with WIN98 system, and the image acquisition and processing are programmed using VC. The following is the program running interface of image acquisition and processing as shown in Figure 5.

3.2 Stable walking of humanoid robot

This control system is used to realize the walking control of DF-1 humanoid robot. Experiments show that this system can complete the dynamic and stable walking of the humanoid machine. Figure 6 is a screenshot of a series of continuous walking.

4 Conclusion

Based on PC/104 embedded computer and C8051F310 chip, the control system of humanoid robot is designed, which realizes the image acquisition and processing of the robot, as well as the stable walking of the robot. PC/104 embedded computer has complete functions, strong computing power and good scalability, and has its unique advantages as a humanoid robot control system. The single chip microcomputer realizes the joint control of the humanoid robot. Due to its limited computing power, it is difficult to realize complex control, so this control system can be used as an experimental and teaching robot.