Design and implementation of hardware circuit for intelligent fire-fighting robot

After more than 50 years of development, artificial intelligence has become a very broad research field and has achieved many remarkable achievements [1]. Artificial intelligence, also known as machine intelligence, is a discipline that studies the mechanism of human intelligence and how to use computers to simulate human intelligent activities. Intelligent robot technology is a high-tech technology formed by integrating multiple disciplines such as computer, cybernetics, mechanism, information and sensor technology, artificial intelligence, bionics, etc. It integrates the development achievements of multiple disciplines and represents the forefront of high technology development [2]. The research on intelligent robots has greatly promoted the progress of artificial intelligence ideas and technologies, and has gradually become a branch field that has attracted much attention. Various intelligent robot competitions have also become a competitive project widely promoted and developed at home and abroad.

The intelligent robot firefighting competition was founded by Trinity College in the United States in 1994. It has now become one of the largest and most popular fully autonomous intelligent robot competitions in the world. The hardware circuit is the core skeleton of the intelligent firefighting robot. Its parameter performance and design rationality directly determine the performance of the intelligent firefighting robot. This paper completes the design and implementation of the hardware circuit of the intelligent firefighting robot based on the ARM9 core [3].

1 Overall design of hardware circuit

The task of the fire-fighting competition is to randomly place candles in one of the rooms in a closed room model to replace the fire source, and the robot is required to find the fire source and extinguish the fire without collision in the shortest possible time.

According to the requirements of the competition and functional needs, the overall structure of the fire-fighting robot is shown in Figure 1, which is mainly composed of modules such as controller, sensor input, and drive output.




2 Analysis and design of main components of hardware circuit

2.1 Embedded system

In order to realize the robot to walk along the prescribed path at high speed and precision, the robot's CPU is required to be able to read the values ​​of multiple sensor ports in real time and quickly, and complete the storage, calculation and output of the values ​​of each port in a short time. Because the embedded microprocessor has a strong support capability for real-time tasks, can complete multiple tasks and has a short interrupt response, a controller with an embedded microprocessor ARM9 as the core is selected in the design process. Its internal Harvard structure can execute 110 million machine instructions per second.

In order to improve the port value reading speed and enable the robot to make rapid judgments on the surrounding environment information, this design sets ADC0~ADC7 (P4.0~P4.7) 8-way data input port on the main chip, which can realize 500,000 data acquisitions per second; in addition, 20-way data input ports are set up, which are connected to the main chip through the ATMEGA816-PC auxiliary microcontroller to read the values ​​of the far-infrared sensor group and the detection port, and can realize 1,000 data acquisitions per second. This design also sets 4-way PWM control signal output ports to drive 4-way high-power DC motors to achieve precise adjustment of the speed; in addition, 7-way Do digital output ports are set up to drive servo motors, buzzers, relays, light-emitting diodes, etc. In order to provide more execution space for large and complex programs, this design additionally sets up 100 KB of data storage (RAM) and 512 KB of program storage (Flash ROM) to store more data and commands.

2.2 Power supply and drive circuit design

(1) Power supply and sampling

circuit The power supply is a key component to ensure the stable and reliable operation of the robot, which directly affects the performance of the robot. Since the motor drive and controller of this robot use two power supplies with different voltage levels, in order to avoid mutual interference between the two power supplies, this robot adopts a dual power supply system: the motor power supply uses a high discharge rate polymer lithium battery with a capacity of 2500 MAH and an operating voltage of 24 V. It can provide a stable power supply current of 40 A, which is 10 times that of ordinary batteries; the controller power supply uses an 8.4 V lithium battery and provides a voltage sampling port for battery detection. The circuit diagram is shown in Figure 2.



In order to obtain the different voltage levels required by the CPU port circuits, this design uses an LM317 T three-terminal regulator and two AMS1117 low-dropout linear voltage regulators, and through their auxiliary circuits, obtains three precise and stable voltages of 5 V, 3.3 V, and 1.8 V; an LED LD1 and a current-limiting resistor R5 are used as power indicator lights to display the status of the power switch; in order to sample the power supply voltage in real time and prevent the lithium battery from being over-discharged or over-charged, the design uses R1 and R2 to divide the voltage and lead out the AD19 port as the power sampling port. (2) DC motor drive circuit Due to the needs of competitive games, the robot must increase its speed as much as possible while avoiding collisions, so it requires a driver with higher power and more sensitive control methods. For this reason, the motor drive power supply voltage used in this paper is 16.8 V and the current is 20 A; a 4-way PWM signal with a duty cycle range of 0 to 95% is used to control the DC motor to achieve precise speed regulation [4]. Since the motor has a large power and is required to be able to achieve bidirectional and adjustable speed operation, this paper designs a half-bridge power MOSFET tube and successfully realizes the control of the motor. As shown in Figure 3, two PWM signals are connected to the MOSFET tube of model IRF2807 through the IR2104 half-bridge driver and the corresponding protection circuit to control the on and off of the power supply and motor connection line to achieve the purpose of controlling the motor speed. When the PWM signal duty cycle is large, the line conduction time is long and the motor speed is high; on the contrary, when the PWM duty cycle is small, the line conduction time is short and the motor speed is low. The four MOSFET tubes are turned on at different times to control the direction of motor rotation: when MSFET tubes 1 and 4 are turned on, motor port 1 is positive and 2 is negative, and the motor rotates forward; when MOSFET tubes 2 and 3 are turned on, motor port 2 is positive and 1 is negative, and the motor rotates reversely. 2.3 Sensors (1) Infrared ranging sensor The infrared ranging sensor [5-6] is the robot’s “visual organ”. By continuously reading its values ​​and making judgments, the robot can determine its location and environment, and determine what command the robot should execute next to avoid collision and walk along the ideal route. According to the specifications of the competition venue, this robot uses SHARP’s GP2D12PSD sensor (hereinafter referred to as PSD sensor), which has an effective ranging range of 10 cm to 80 cm. The principle is shown in Figure 4 (a).

The sensor uses the principle of triangulation. As shown in Figure 4 (b), the infrared light emitting diode emits an infrared beam. When the infrared beam encounters an obstacle in front, a part of it is reflected back and focused on the linear charge coupled device CCD (Charge Coupled Device) behind it through the lens. According to the position where the infrared light is focused on the CCD, the reflection angle of the light can be known, and the distance of the object can be further calculated. Since the output voltage of the PSD sensor and the actual distance are nonlinear, the conversion approximate formula can be obtained through linear interpolation.

According to the needs of the competition, the robot should be able to measure the distance of obstacles in different directions. In theory, infrared ranging sensors should be set in 8 directions. Under the premise of meeting the requirements of the competition and considering economy, this design uses 6 infrared ranging sensors, and their placement positions are shown in Figure 4 (c). The relationship between the position of the robot and the wall can be determined more accurately through one or more sensor values. For example, when the values ​​of the front sensor and the left front sensor are both large (the distance is very small), it means that the robot is in a corner, with walls in front and on the left. At this time, the right turn command can be executed to get out of the corner.

(2) Far-infrared flame sensor group

In order to complete the fire-fighting task, the robot must be able to determine the approximate location of the flame and determine whether the flame has been extinguished. This paper designs two far-infrared flame sensor groups consisting of 28 infrared receiving tubes. Each front and rear position has 14 infrared receiving tubes. Every two are connected in parallel and point to the same direction. The two sensor groups point to 14 directions in total, which can cover a 360° range. As shown in Figure 5 (a), the 14 ports are connected to the ATMEGA8-16PC microcontroller through the CD4051 eight-way conversion switch, where SCK, MISO, and MOSI are bit selection ports. In addition, this design can also compare the 14-way read data to determine its maximum and minimum values ​​and the corresponding port values, which facilitates the determination of the fire source. By



comparing the different port values ​​of the far-infrared sensor group, the relative position of the robot and the fire source can also be determined to determine the forward direction and complete the light-approaching action. When the robot is in the relative position to the fire source as shown in Figure 5 (b), the values ​​of port 2 and port 4 can be read and subtracted. If the value of port 2 is greater than that of port 4 (indicating that port 2 is closer to the fire source), a left turn command is executed to make the difference within a certain range, and then a straight-line command is executed to approach the fire source.

(3) Ground grayscale sensor

According to the competition rules, the robot's starting position is a white circle with a diameter of 30 cm. There is a 3 cm wide white line at the entrance of each room, and the rest of the ground is black. The robot's start and stop and the sign of entering the room all rely on the judgment of the ground grayscale, so a sensor that can respond to the intensity of light reflected from the ground is required. This machine uses a pair of ground grayscale sensors placed on the bases at the front and rear ends. The darker the ground color, the larger the value, and the lighter the ground color, the smaller the value.

As shown in Figure 6, the ground grayscale sensor illuminates the ground through the light-emitting diode LED, and the reflected light of the ground is received by the phototransistor. When the ground color is black, the reflected light is relatively weak, then the base current of the phototransistor is smaller, the collector current is also smaller, the voltage value of port 1 is higher, and its measured value is larger; on the contrary, when the ground is white, the reflected light is stronger, the collector current is larger, the voltage value of port 1 is smaller, and the measured value is also smaller.



This paper studies and designs an intelligent fire-fighting robot based on the ARM9 embedded system, which has the following five innovative points: (1) The embedded system kernel is adopted, which greatly improves the robot's ability to process signals; (2) The introduction of the dual power supply system makes the operation of the robot more stable and reliable; (3) The use of PWM signal to control the high-power DC motor has greatly improved the speed and accuracy; (4) By reasonably selecting the number and placement of PSD ranging sensors, it can meet the requirements of the competition and save costs; (5) The far-infrared flame sensor group designed in this paper has well completed the task of accurately locating the fire source and improved the reliability and rapidity of fire extinguishing.

Actual tests have proved that the robot designed in this paper can complete the competition tasks well, and has greatly improved its reliability and speed, and has strong application value.