Authors: Wang Chunbao; Jiang Liangzhong; Lin Zhuohua; Li Chaoqing
EOD robots are a type of handling robots. Many handling robots imitate the actions of humans and are used to help or partially replace humans to perform heavy, dangerous, and repetitive tasks. EOD robots imitate human walking and grasping objects, and can be used to remove explosives at the scene. The mechanical part of the EOD robot includes a walking trolley and a robotic arm.
1 The structure of the control system of the EOD robot
The control system of the EOD robot no longer adopts the PLC control method used by the original robot, but is based on the embedded industrial computer and adopts an open component library based on the Linux platform. The operation speed is extremely fast, which can reach the microsecond level. The control system of the EOD robot consists of three parts, namely the sensor part, the robot body, and the operation console. The sensor part includes wide access and multi-information fusion device. Wide access and multi-information fusion are based on ARM9 system, which consists of A/D with 12-bit precision, 32-channel I/O, and RS232 communication interface, which can integrate various sensor signals; the robot body includes an embedded control computer and each joint servo motor , can drive each axis (with compensation), distribute the motion of each axis, process and control various state quantities; the operation console part includes the main control PC, instruction decoder, image decoder, wireless receiver and transmitter, etc. The image information and obstacle information obtained by the CCD camera and the ultrasonic sensor are wirelessly transmitted to the main control PC. The main control PC uses image recognition and analysis technology to obtain the shape and position information of the object, and at the same time displays the three-dimensional image information of the target and obstacles on the computer screen. After manually specifying the target, the main control PC is calculated by artificial intelligence (pattern recognition, path planning, trajectory collision avoidance, etc.); the control instructions are wirelessly transmitted to the embedded control computer of the robot body to automatically control the movement of the robot. The basic structure is shown in Figure 1:
Figure 1 Basic structure of the control system of the EOD robot
2 The walking control of the trolley
Manually through the control lever (or button), under the instruction of the car guidance image (visual signal), complete the functions of forward, backward, left turn, right turn and in-situ rotation until the gripper of the manipulator can grab suspicious explosives up to the position. The balance of the robot is obtained by using the dynamic algorithm of the whole machine, and a variety of data will be obtained to control the spatial posture of the robot arm so as to achieve the balance of the whole machine without overturning. For example, when the trolley moves up on the slope, the robot arm leans forward to maintain the balance of the whole machine. The car needs to avoid obstacles that may be encountered during the movement. Through environmental sensors such as CCD cameras, ultrasonic sensors, etc., the size of obstacles and environmental information are obtained to achieve obstacle avoidance, which is the so-called path planning (as shown in Figure 2).
Figure 2 Schematic diagram of the path planning of the EOD robot
3 Motion control of the manipulator
The motion control of the manipulator (as shown in Figure 3) is a multi-level control system: the AI level at the top is the artificial intelligence level. It uses sensors such as CCD cameras and ultrasonic waves to obtain the outline and distance of obstacles to avoid obstacles to avoid collisions, which is “trajectory planning” (see Figure 4).
Figure 3 Multi-level motion control system of manipulator
Figure 4 Schematic diagram of robot trajectory planning
The control in the middle layer is the most critical part of the gripping motion control of the manipulator. It’s actually a tween. The motion control of the manipulator can be divided into attitude control and hand position control. The control of waist, big arm, forearm and claw is solved by the main control computer through the kinematic displacement problem of the manipulator to solve the inverse motion to obtain the positions of the waist, forearm, forearm and claw, and then to give each joint The position command is transmitted to the embedded system control computer of the above-mentioned joint to complete the position control with the speed feedback loop. The motion speed of the four joints is determined by the trajectory planning of the manipulator, so as to avoid the serpentine motion of the manipulator. Thanks to tactile sensors, the gripper can be controlled automatically, or it can be finely “softly controlled” under the guidance of image signals. When grabbing suspicious explosives, control the gripping force of the claw to be about greater than the gravity of the suspicious explosives falling off, so as not to squeeze the suspicious explosives with excessive force, so as to complete the action specified by the experts.
The bottom layer is the control level. Both the position controller and the speed regulator use mature PID (including P, PI, PD) control. When PID control selects appropriate parameters, it has certain robustness, simplicity and practicality. When PID cannot achieve effective control, it can be improved or combined with other methods, such as feedforward control. The motion control includes the motion control of each joint of the manipulator and the walking motion. There are 13 preliminary statistical control loops, as shown in Figure 5.
Figure 5 Schematic diagram of automatic control
The embedded control computer sets the position and speed of each servo mechanism according to the position and speed signals from the upper computer, and controls the servo motor according to the control law (such as PID), so that the actual position can be tracked or moved to the desired speed. target location.
Robot’s hand-eye coordination system: The robot’s hand-eye coordination system is one of the biggest highlights of this system. Since the tween algorithm needs to know the base coordinates of the target and the robot (usually the waist joint base of the robot), the relative space coordinates can be solved. The target object of this system is in an unknown environment. This relative coordinate has to be solved due to the hand-eye coordination system. The EOD robot manipulator is equipped with a binocular system, and a “moment measurement triangle” can be formed by using the binocular system and the target. The principle is shown in Figure 6. The distance error obtained by the moment-measuring triangle is relatively large, and it is necessary to use the “light spot” of the laser ranging to assist the moment measurement, so as to obtain the spatial coordinates of the coordinate object. As the robot’s binocular approaches the target, the relative coordinate accuracy between the target and the base coordinate improves. After several binocular approaches to the target, the manipulator finally achieves the target accurately grasped.
Figure 6 Schematic diagram of the moment measuring triangle
When the robot fails to grab automatically, the operator can use the manual system to complete the grab. This is a function that any product must have after automatic failure. The image information obtained by the robot’s binoculars is processed to form a monitoring image on the display of the console, so that the operator can click and select suspicious objects among multiple objects with the mouse.
4 Communication system of the control system
Serial wireless communication is adopted between the robot and the main control computer of the on-site operation console. The on-site master computer and remote monitoring center use CDMA communication technology for wireless image transmission. The status of on-site objects and robots can be observed at any point in the communication system.
In laboratory and industrial applications, limited by the channel cost, the serial port is often used as the preferred data transmission channel between the computer and external serial devices, and due to the convenience of serial communication, many devices and computers can use the serial port to communicate with peripherals. Control, detection and serial communication have increasingly become a very important means for the computer to communicate with the peripherals and to obtain the monitoring data collected by the peripherals. In order to solve the limitation of wired communication, considering the characteristics of robot self-planning, self-organization, strong self-adaptation and uncertain environment, wireless communication is an ideal way to realize the communication between robot and computer. Therefore, the use of wireless data transmission module for data signal transmission is an important part of the EOD robot. The wireless data transmission module is small in size and easy to use, and can control the robot from a long distance to effectively protect the safety of operators.
The so-called CDMA (Code Division Multiple Access) is a carrier modulation and multiple access technology based on extended communication systems. CDMA has many advantages, some of which are inherent in the expansion system, and the other part is brought about by software switching and power control. CDMA communication networks are composed of spread spectrum networks, multiple access, cellular networks, and frequency reuse. It is a combination of three-dimensional signal processing technology in the frequency domain, time domain and code domain, so it has good anti-interference performance, anti-multipath fading, and high security performance.
5 Conclusion and Outlook
The innovation of this paper is that the control system of the EOD robot is an open control system. The control system successfully combines binocular vision ranging and intelligent control algorithms to realize the intelligent and networked functions of the robot. The applicability of the design is very wide, as long as the appropriate modification is made according to the needs, the control system suitable for other functional robots can be designed, which has a high reference value.
Through the research and development of the system, the automatic control level of the EOD robot will be improved, the operability and reliability of the EOD robot will be improved, and it will be of great help to ensure public safety and improve the EOD efficiency.
Responsible editor: gt