This paper presents the development of a line follower wheeled mobile robot. In this project, ARM cortex-3 based microcontroller is chosen as the main controller to react towards the data received from infrared line sensors to give fast, smooth, accurate and safe movement in partially structured environment. A dynamic PID control algorithm has been proposed to improve the navigation reliability of the wheeled mobile robot which uses differential drive locomotion system. The experimental results show that the dynamic PID algorithm can be performed under the system real-time requirements. Keywords – embedded system, wheeled mobile robot, PID control algorithm.

IntroductionEmbedded system includes many areas of knowledge, microcontroller hardware and software, interfacing technologies, automatic control theory, and sensor technologies etc. To speed up the learning process and motivate students to learn actively, the project-based learning approach may be applied in the embedded system design laboratory [1-4]. The low-cost wheeled mobile robot building, which is proposed in this paper, serves as a good example on which students can learn embedded system design skills. It covers not only common embedded system peripherals, but also energy control and real-time control firmware implementation. The process of the construction of wheeled mobile robot can give students the idea that hardware circuits and software algorithms are both mandatory for a successful embedded system design. The competition between student groups in the racing contest can also encourage them to explore in depth the skills acquired in this laboratory as well as give them lots of fun [5-7].

The remainder of this paper is organized as follows: The line follower robot structure and architecture issues and challenges along with their technical issues and problems are discussed in section 2. Programming details will be explained in section 3. Section 4 describes the integration of the complete system. LA„A±ne follower wheeled MobA„A±le Robot structureGenerally, the line follower robot is one of the self-acting wheeled mobile mechanisms that follow a line drawn on the floor. The route can be a visible black line on a white surface or vice versa. The simple operations of the wheeled mobile line follower robot are shown below.

Taking the line position data with optical sensors attached at the front end of the mobile robot. Most are using more than a few numbers of IR photo-reflectors. Therefore, the line sensing procedure needs high resolution and high robustness. Steering the wheeled mobile robot to track the line with any direction-finding mechanism. This is just a servo maneuver; actually, any phase recompense will be required to become stable following motion by applying digital PID filter or any similar servo algorithm.

Monitoring the speed according to the path complaint. The speed is restricted during passing a turn due to the friction of the tire with the floor. From actually building the robot platform, to setting up, programming, and hardware or software fine tuning, everything needs to be taken into account when building a differential wheeled mobile robot. A mobile robot can be considered fundamentally as a combination of five main portions and subsystems.

Chassis and body. Sensors and signal processing circuits. Microcontroller and interface circuits. Motor drivers Actuators (Motors and wheels) The Chassis and BodyThe Chassis would be the first part of a robot’s body. It is designed to handle all of the other components, transmission mechanisms, electronics and battery. It needs to be sufficiently large and provide adequate fixtures to furnish all necessary parts, as well as sturdy enough to cope with the weight of the parts along with additional loads which can appear in dynamic conditions such as vibrations, shocks or chassis torsion and actuators torque.

There are some good materials for designing robots such as plastic, aluminum and carbon-composites. We must pay attention to the resistance, weight and mechanical ability for choosing one of them. In the designed robot, printed circuit board (PCB) has been used for chassis because of its lightweight and being strong enough for robot project. All components can be installed on the PCB to minimize the weight. It is noted that the performance is much more valuable than other issues.

Sensors and Signal Processing CircuitLine follower robot uses Infrared Ray (IR) sensors to find the path and direction. IR sensors include an infrared transmitter and infrared receiver pair. IR sensors are often used to identify white and black surfaces. White surfaces effectively reflect well, but black surfaces reflect poorly. Hence, the distance between sensors and ground surface is important, and it is more valuable that how we put sensors near to each other. The distance between sensors and ground surface must be 2 to 10 mm, and the distance between each sensor is dependent on the line width.

In the designed robot, we have used eight sensors, and they have a suitable distance between each other. If the line width is narrow, the distance between sensors must be reduced; otherwise, while curving the line, the robot will not be turned on time. Generally, the received signals from the sensors are analog and must be converted to the digital form. Therefore, the designed signal processing circuit can send the sensors’ signals to the microcontroller directly. MicrocontrollerWe have used the TI Stellaris microcontroller LM3S811 in robot project.

The LM3S811 microcontroller has a Reduced Instruction Set Coding (RISC) core. Internal oscillators, timers, UART, USB, SPI, pull-up resistors, pulse width modulation, ADC, analog comparator and watch-dog timers are some of the features [8]. With on-chip in-system programmable Flash and SRAM, the LM3S811 is a perfect choice in order to optimize cost. Motor DriversA well-known and suitable motor driver is IC L298 which can be used to control two motors.

It is a high voltage, high-current dual full-bridge driver designed to accept standard TTL logic levels and drive inductive loads such as DC and stepping motors [9]. Two enable inputs are provided to enable or disable the device independently of the input signals. L298 has 2 amperes per channel current capacity and it can support up to 45 volts for outputting. Moreover, L298 works well up to 16 volts without any heat sink. The Actuators (Motors and Wheels)There are many kinds of motors and wheels. Our choice depends on the robot function, power, speed, and precision.

Actually, it is better to use gearbox motors instead of common DC motors because it has gears and an axle and its speed does not change towards the top of a hill or downhill. Motors are rated to operate at 1700 rpm at 7 volt nominal voltage. It is better to use wheels for line follower robots, instead of a tank system. We can use three wheels.

Two of them are joined to the motors and installed at the rear of the robot and the other wheel is free and installed in front of the robot as a passive caster. To get better maneuver, robot uses two motors and two wheels on the rear and a free wheel on the front. The power supply is 7.6 V with a regulator. The designed robot has eight infrared sensors on the front bottom for detecting the line.

Arm based microcontroller Stellaris and driver L298 were used to control direction and speed of motors. General view of the line follower robot that we built is shown in Fig. 1. The robot is controlled by the microcontroller.

It performs the change in the motor direction by sending an appropriate signal to the driver IC according to the received signals from the sensors. Real Time Task SchedulingWe built a light-weighted and high-speed robot because points are awarded based upon the distance covered and the speed of the overall robot. Therefore, we used two high speed motors and a highly sensitive signal conditioning circuit. The body weight and wheels’ radius have effects on the speed, too. The weight of the designed robot is around 300 gr. and it could be lighter.

The photograph of the top and bottom views of the designed robot is shown in Fig. 1. The microcontroller sends instructions to the driver after processing the data received from sensors. The driver powers the motors according to the inputs.

Actually the driver supplies positive voltage to one of the motor pins and negative voltage to the other. There are five states of movement: To move forward; both of the motors are turned on and rotate forward simultaneously. To move left; the right motor is turned on and the left motor is turned off. To move right; the left motor is turned on and the right motor is turned off. To move left fast; the right motor rotates forward and the left motor rotates backward.

To move right fast; the left motor rotates forward and the right motor rotates backward. Most embedded system applications need to react to the inputs or environment changes in real time, which means that the accuracy of computations is as important as their timelines. Furthermore, digital control algorithms need a fixed sampling time interval for measuring inputs and delivering output commands. Therefore, the idea of applying interrupts for task scheduling is introduced in this work. IMAG0388 (a) IMAG0392 (b) Figure 1 – Images show (a) top, (b) bottom views of the built line follower robot.

The Quadratic Line-Detection AlgorithmA better way of detecting the line position, compared to the other simple line-following robots, by using a quadratic interpolation technique is introduced. Eight reflective optical sensors were used, and the coordinate of the leftmost sensor was 0. To find out the correct position of the black line, we had to locate three consecutive sensors with higher output readings than the other five sensors as shown in Fig. 2.

Assume that the coordinates of these 3 sensors are x1, x1+1, and x1+2, and the true shape of the sensor output values are in the range of [x1, x1+2] which can be approximated by a quadratic curve. One can then find the following relationships between the coordinates of the sensors and the output values: (1) (2) (3) The coordinate value, at which the output value of the quadratic curve is the maximum, is considered as the true position of the line. By using the basic calculus, one would know that the coordinate value is: (4) (5) (6) It is assumed that the coordinate for the center position of the line-following robot is 0. Therefore, the error e between the line position and the center position of the robot is e a‚¬A?a‚¬A 0a‚¬A­a‚¬A x a‚¬A?a‚¬A a‚¬A­x (7) Figure 2 – The line detection algorithm via quadratic interpolation.

PID Tracking Control AlgorithmThe popular proportional-integral-derivative (PID) controller was introduced in this project to make the robot follow the racing track. The error between the center of the sensors and the track to be followed was then processed by the PID controller to generate velocity commands for the right and left wheels. First, the controller calculates the current position and then calculates the error established on the current situation. It will then send commands the motors to give a rigid turn, if the error is extraordinary or a minor turn, if the error is small. Basically, the amount of the turn given will be proportional to the error.

Of course this is a consequence of the proportional control. Even after this, if the error does not decline approximately to zero, the controller will then growth the degree of the turn further and further over time till the robot centers over the line. This is the result of the integral control. In the process of centering over the line, the robot may overshoot the target position and move to the other side of the line where the above process is followed again.

Thus, the robot may keep oscillating about the line in order to center over the line. To reduce the oscillating effect over time, the derivative control is used. The proportional term is only a gain amplifier, and the derivative term is applied in order to improve the response to disturbance, and also to compensate for phase lag at the controlled object. Pseudo Code for the PID Controller; Kp = 10 Ki = 1 Kd = 100 offset = 45 ! Initialize the variables Tp = 50 integral = 0 ! the place where integral value will be stored lastError = 0 ! place where last error value will be stored derivative = 0 ! place where derivative value will be stored Loop forever LightValue = read sensors ! read sensors.

error = -x ! calculate the error using equation (7). integral = integral + error ! calculate the integral derivative = error – lastError ! calculate the derivative Turn = Kp*error + Ki*integral + Kd*derivative powerA = Tp + Turn ! power level for motor A powerB = Tp – Turn ! power level for motor B MOTOR A direction=forward power=PowerA MOTOR B direction=forward power=PowerB lastError = error ! save the current error end loop forever ! do it again. PID controller requires the Kp, Ki and Kd factors to be set to match wheeled line follower robot’s characteristics and these values depends on robot structures, actuators, sensors and other electronic components’ characteristics. There is no equation given in the literature to calculate Kp, Ki and Kd factors. It requires experimental trial and error technique until you get the favorite behavior.

We defined these factors according to following guidelines; Start with low speed and setting values of Kp, Ki and Kd to 0. Then, try setting Kp to a value of 1 and observe the robot. The goal is to get the robot to follow the line even if it is extremely wobbly. If the robot overshoots and misses the line, decrease the value of Kp.

If the robot cannot navigate a turn or seems listless, increment the Kp value with small steps. Once the mobile robot is able to follow the path, set Kd value to 1 and then try growing this value until you see less shake. Once the robot is fairly stable at following the line, assign a value of .5 to 1.

0 to Ki. If the Ki value is extraordinary, the robot will shake left and right rapidly. If it is too low, you won’t see any perceivable alteration. Since integral is increasing, the Ki value has a substantial impact. You may continue to retuning process with adjusting Ki by .01 increments.

Once the mobile robot is tracking the line with reasonable accuracy, you can increase the speed and see if it is still able to track the line. Speed disturbs the PID controller and will require rearranging as the speed fluctuations. Results And DiscusionA line following robot is programed with simple (on/off) control as a comparison purpose in evaluating the performance of the dynamic algorithm controlled robot. The results of the experiment are summarized in Table-1. From the data in the table, it can be observed that dynamic PID algorithm controlled robot has better performance in every criteria listed in the table compared to simple (on/off) control robot. The dynamic algorithm controlled robot has higher velocity, consumes less time to complete one whole circuit, tracks the line smoother and has lower tendency to astray from line compared to uncontrolled robot.

Therefore this system can be used in training undergraduate students on dynamic PID algorithm control system, its application and implementation in the real world and the advantages that it offers. Fig. 3 shows the designed robot during race pits test. Figure 3 – The designed robot on the race pits. Table 1- Experimental result for Line Following Robot. Criteria Dynamic PID algorithm Simple (on/off) Time to complete one whole circuit 47.

6s 71.4s Line tracking Smooth Not so smooth Velocity 0.2m/s 0.14m/s Tendency to astray from line Low High ConclusionThe designed wheeled line follower mobile robot has eight infrared sensors on the bottom for detecting the line. The controller board includes Stellaris LM3S811 micro-controller and the motor driver L298 which were used to control the direction and the speed of motors.

The proposed dynamic PID algorithm derives the line follower locomotion by adequately combining the information from sensor module. Experimental results show that the proposed algorithm can successfully achieve target following in various scenarios, including straight line and circular motion, sharp-turn motion and S-shape line tracking. We are working currently to develop a more sophisticated algorithm which can perform faster line tracking with less energy consumption.