Abstract : This paper focuses on line-following strategies, taking a smart car using dual-row sensors as an example, and presents optimized line-following strategies for different road conditions such as straight lines, sharp curves, and S-curves. Experiments demonstrate that this strategy closely integrates the characteristics of dual-row infrared sensors, leveraging their advantages to enable the smart car to prioritize stability while pursuing maximum speed. It exhibits strong adaptability and performs well on various tracks. Keywords : Dual-row; Sensor; Path recognition; Line-following strategy; Advantages of dual-row sensors Currently, most smart cars use single-row sensors for road detection. This method obtains limited road information and cannot effectively distinguish between the smart car's state and the road conditions, causing control difficulties. To compensate for this, a large-lookahead single-row sensor road detection method has been developed. This method detects at a greater distance and can determine the road direction earlier, mitigating the low detection accuracy to some extent. However, it still cannot effectively distinguish between the smart car's state and the road conditions. The racing car model can use cameras or sensors to detect road information. Our car model uses a dual-row infrared tracking method. The large-look dual-row sensors can obtain more track information, allowing for earlier strategic processing and a better driving trajectory. This is an alternative to using complex camera solutions. It can achieve stable control and smooth acceleration on straightaways; it can advance along small curves in S-curves, reducing the number of route and servo adjustments. It can achieve early turning and inside-cutting effects in sharp curves. Especially in cornering, by predicting the curve together with the front and rear sensors, it can extend the physical recognition distance, thus making earlier actions and reducing the negative impact of short detection distances, achieving the above effects. Sensor Array Layout Figure 1 only shows the sensor positions using the receiving tube. Figure 1 Sensor Array Layout Explanation: • The front sensors extend further; if the car's center deviates from the black line, a larger offset will be generated on the front sensors. • The rear sensors extend closer; if the car's center deviates from the black line, a smaller offset will be generated on the rear sensors. • The car is controlled by utilizing the different sensitivities of the front and rear sensors when the car deviates. • In order to make the front and rear rows more clearly differentiated and collect information from further away, we tilt the front row sensors at an angle of about 45 degrees, so that the forward look-ahead distance of the front row is greater and the advantages and characteristics of the front row can be better reflected. Straight road recognition method and control strategy Straight road recognition method (1) Using this method to arrange dual-row infrared sensors, there are five physical methods for judging straight roads. The timing of each method is listed in the table. First straight road situation (Figure 2) Figure 2 The first straight road situation is the combination of the front and rear sensors detecting the black line when exiting the left turn after a large left turn. Applicable to left 90° turns and 180° turns. The exit information is obtained in advance, the servo turns to the left by a small angle, and accelerates at this time to compensate for the lack of forward look-ahead. When the S-curve of the track appears, this situation does not meet the second recognition method of straight road, so it will not accelerate. The second straight-line scenario (Figure 3) [align=center] Figure 3 The second straight-line scenario[/align] This scenario is a reconfirmation of the first scenario. After making a wide left turn and passing through the first scenario, experiencing this scenario confirms that the road ahead is straight, and the car's acceleration is further enhanced. The control program switches from the cornering program to the straight-line stability program. The third straight-line scenario (Figure 4) [align=center] Figure 4 The third straight-line scenario[/align] Straight-line stability control is adopted at this time. Since the first two scenarios have been clearly identified as straight roads, this scenario only increases the success rate of straight-line recognition. The fourth straight-line scenario (Figure 5) [align=center] Figure 5 The fourth straight-line scenario[/align] Similar to the second scenario, this scenario is a reconfirmation of the fifth scenario. After making a wide right turn and passing through the fifth scenario, experiencing this scenario confirms that the road ahead is straight, and the car's acceleration is further enhanced. The control program switches from the cornering program to the straight-line stability program. Fifth straight road situation (Figure 6) [align=center] Figure 6 Fifth straight road situation[/align] After a large right turn, the most likely combination of front and rear sensors detecting the black line when exiting the turn. Applicable to 90° and 180° right turns. Obtaining exit information in advance, the servo turns to the right by a small angle and accelerates at this time to compensate for insufficient foresight. When the S-curve of the track appears, the second identification method of the straight road is not satisfied, so there is no acceleration. (2) Straight road identification, the program assists in confirming that after entering the curve, as the car moves, there will be oscillation, which may cause the above 5 situations to not be satisfied when exiting the curve. In order to improve the success rate of straight road identification, a second straight road discrimination method is added. Both work at the same time, and after satisfying the first one, it is confirmed to be a straight road after a maximum of 15ms. The program is executed in a loop, and our program execution frequency is 2KHz. A timer interrupt (15ms) is used to count the three sensors (numbered 3, 4, and 5) in the front row using three counters. Each time the program executes, if one of the sensors detects a black line, the corresponding counter increments by 1. Calculations show that the maximum count within 15ms is 31. We set a maximum count value; if the required count is reached within 15ms, it's considered a straight road, the straight road program is switched, and the counter is reset to zero. If the required count is not reached within 15ms, the counter is reset to zero, and the counting restarts. For example, if the car travels at 2m/s and moves 3cm, we only need to determine if it's a straight road within 2-2.5cm. Therefore, a maximum count of 20-25 is considered a straight road, and the curve program is exited. Of course, a more stringent method can be used by adjusting the timer interrupt duration and count value. This condition is always met after entering a straight road, so it serves as a supplement to the first straight road detection method, ensuring stable and reliable straight road identification. Straight-Line Stability Control Strategy After exiting a curve, the car oscillates due to the sluggish response of the servo motor before stabilizing. To minimize oscillations as early as possible, the following method is used to control the car's movement after exiting the curve: A flag is set in the curve strategy. Upon entering the straight-line program, the flag is identified, and the formula for controlling the servo motor's steering is modified. The formula is: q = K1q1 + K2q2; where q is the final control quantity sent to the servo motor, q1 is the return angle value from the front photoelectric sensor, and q2 is the return angle value from the rear infrared sensor. K1 and K2 are the weighted ratio values ​​of the front and rear sensors, respectively. Normally, K1 and K2 are 1, and their values ​​are changed when necessary. When the car enters the straight section from the curve and successfully identifies the straight section, the value of K1 is decreased. Because the rear sensors are very close to the car's front wheels (steering wheels), when the car's center deviates from the black line, there will be no significant lateral displacement (relative to the front sensors) in the rear sensor's position. Therefore, the number of servo motor adjustments on the straight section is significantly reduced, resulting in better straight-line stability. Meanwhile, based on the combination of different sensors in the front and rear rows, different turning strategies are given (represented in a list format in the program), further improving the stability control capability of the straight line. References : 1. Shao Beibei, Online Development Methods for Embedded Applications of Single-Chip Microcomputers [M], Tsinghua University, 2004. 2. Zhuo Qing et al., Learning to Make Intelligent Cars, Beijing University of Aeronautics and Astronautics Press, 2007.3.