Fun Projects for your LEGO^® MINDSTORMS^® NXT!

Note: The gray skis and the light sensor are on the front end of the robot. The NXT display should be right side up for reading when you are behind the robot.

When building a robot that depends on sliding on something (the gray front 'skis' in this project) to turn, it is important to consider the weight balance of the robot. You want to have most of the robot's weight over the drive wheels, and just enough weight over the sliding part to keep the robot stable and avoid tipping over. If too much weight is over the sliding part, the robot will struggle to turn, might get caught up and stall, or the drive wheels might start slipping.

This Line Follower robot is designed to have almost all of its weight over the drive wheels. Using a small scale in the experiment shown below, we find that the robot weights a total of 622 grams, with 560 grams over the back wheels. This means that the robot has 90% (560/622) of the weight over the wheels and only 10% of the weight over the front skis. This is good for line following at slow speeds, but note that it could cause the robot to pop a wheelie if a lot of power was suddenly given to both wheels going forward (see the Dragster).

There are three programs provided for the Line Follower:

The LineFollow2 program is a basic 'Two State' line follower that uses a simple 'zig-zag' method of line following where the robot is constantly turning back and forth as it sees either side of the color boundary. The robot is always either turning left or right (the two states), so it is never actually straight, even when the line is straight. Before using this program, you will want to calibrate the sensor (on the robot) for the lighting conditions of the surface and line you are using (see below for instructions). To get the best start for the line following, start the robot with the sensor near the right edge of the line.

The LineFollow5 program uses a somewhat more complex 'Five State' method of following the line with the goal of going straighter and smoother (and therefore faster). When the robot gets aligned with the color boundary that it is trying to follow, the robot will go straight as far as it can and make small corrections left and right to try to stay on the boundary without zigzagging as much. See the program comments for details. Before using this program, you will want to calibrate the sensor (on the robot) for the lighting conditions of the surface and line you are using (see below for instructions). To get the best start for the line following, start the robot with the sensor near the right edge of the line.

The LineFollowPro program uses a 'Proportional' line following strategy, which involves several math calculations to calculate an amount of steering correction that is proportional to the distance off of the desired path (the right edge of the line) that the robot thinks it is. This provides for a smooth and continuous transition from going straight, to small corrections, all the way to large corrections, which allows the robot to drive smoother and faster. In addition, the LineFollowPro program starts with an automatic calibration sequence, where the robot scans the line and the surface to the right of the line to automatically determine the minimum and maximum expected brightness values. This eliminates the need to calibrate the light sensor. The LineFollowPro program should be started with the light sensor directly over the center of the line.

Before line following using either the LineFollow2 or LineFollow5 program, you should 'calibrate' the light sensor to the actual conditions expected. With the sensor mounted on the robot where it will be used, and the robot placed over the actual line it will be following, a calibration process takes two light sensor readings, one directly over the line (minimum = darkest), and one over the surface away from the line (maximum = brightest). The NXT will then adjust the numbers reported by the light sensor so that the minimum value expected will be reported as 0 and the maximum as 100. The results of light sensor calibration are remembered until you calibrate again, even after the NXT is turned off (the results are stored in a data file in the NXT's memory), so you only need to calibrate once each time your conditions change. Then you can run your robot as many times as you want without needing to calibrate before each run.

Calibration allows you to make make certain assumptions about the light readings to simplify the program. In particular, the LineFollow2 program assumes that the edge of the line, over which the sensor sees about half line and half surface, will report a brightness of 50. The LineFollow5 program assumes a midpoint of 50 and also assumes that the total range of light readings will range from 0 to 100.

In addition to allowing a program that uses the light sensor to be simpler, calibration also makes it possible to adapt the robot to different lighting conditions without changing the program. For example, if you want to run the robot over a different colored surface (or the room lighting changes), you would simply have to re-calibrate the sensor before running the same line follower program on the new surface.

How to Calibrate the Light Sensor

There are two ways to calibrate the light sensor: using the method built in to the NXT programming software, or using a (somewhat friendlier) calibration program provided below.

To use the method built into the NXT software, connect your NXT to your PC (with the NXT cable or Bluetooth), then select the Tools -> Calibrate Sensors menu item, choose Light Sensor, and choose the correct sensor port (port 3 is used by these line following programs) and press the Calibrate button. At this point the software will download a program called 'Calibrate' to the NXT and run it. Place your robot over the surface to be used and use the screen prompts to sample the two values (see Taking the Two Calibration Readings below).

A similar but somewhat friendlier calibration program is provided here:

The Calibrate 3 program will prompt you to sample two light readings for a light sensor on port 3, one for 'Black' and one for 'White' (see Taking the Two Calibration Readings below). Note that 'White' just means the lightest part of the surface being run on. If your surface is tan colored, for example, just sample that for 'White' (and similar for 'Black'). The program then follows up with a test of light readings that continuously reports the calibrated brightness seen, so that you can easily see and verify the results of your calibration (after calibration, 'Black' should report very close to 0 and 'White' very close to 100).

The View Light 3 program can be used as a measurement tool to simply display the calibrated light readings currently seen by a light sensor on port 3. This is useful because the built-in 'View Reflected Light' feature of the NXT reports uncalibrated readings and will therefore not match the values you want to use in your programs.

The Del Calibrate program will delete the stored light sensor calibration from the NXT's memory and restore the 'uncalibrated' state, in case you want to use that or experiment with the difference.

Taking the Two Calibration Readings

Using either the built-in NXT Calibrate program, or the Calibrate 3 program provided above, you will use the on-screen prompts to sample two light readings, with the robot on the actual surface to be used and the sensor in its actual position on the robot.

• The large black oval on the NXT Test Pad is interrupted by the 'Start' on one end. What do you think will happen to the robot when it gets there? Does it depend on which direction around the track it is going? Does the same thing happen every time?

• The line following programs provided here use the motors at only about half power, in order to stay in control and not miss the turns in the test track. If it goes too fast, the robot may cross over to the other side of the black line and get confused. Thinking about how the program works, can you explain why this confuses it? Madhubala serial cast. What will happen afterwards?

• The LineFollow5 program breaks the driving and steering into five different conditions, depending on how far away from the color boundary the robot thinks it is. By adjusting what is done in each of these cases, can you get this robot or another robot design to follow the line smoother or faster? Try timing it once around the oval then see if you can improve on this time.

• The LineFollowPro program includes two variables, Power and Gain, written right at the beginning, that are adjustable. The Power variable will affect the overall speed of the robot, and the Gain affects how sensitive the steering response is. Try modifying the values written to these variables to better suit the line you are following to see if you can get better results (see comments in the program for more info).

• See if you can improve the robot design or make a new design that will perform better (follow faster or smoother). For example, the distance that the light sensor is mounted in front of the drive wheels affects how the line following, because it changes the sensor's response to the robot's steering. Experiment with different sensor positions and see if you can get better results. This Line Follower design might be able to go faster and/or smoother if the light sensor was closer to the wheels (not so far out in front). Making the wheel track wider left to right would probably help too. When you make adjustments to the robot, experiment with different values of the Power and Gain variables in the LineFollowPro program to adapt for it. Can you make the power higher? Does adjusting the Gain higher or lower help?

• The standard NXT kit comes with only one light sensor, but if you have two light sensors, then it is possible to make a different kind of line following strategy that is sometimes more reliable than with one sensor. If you have two light sensors, try placing them on either side of the line so that they both see white when the robot is centered over the line. Then an example 'Five State' strategy would be:

  1. When both sensors see white (e.g. > 75%), go straight.

  2. When the left sensor sees some black (e.g. 25-75%), make a slight left turn.

  3. When the left sensor sees all black (e.g. Nanny mcphee film. < 25%), make a hard left turn.

  4. When the right sensor sees some black, make a slight right turn.

  5. When the right sensor sees all black, make a hard right turn.

  The programming will be trickier because you must test both sensors. Note that for calibration, you only need to calibrate one sensor, and this same calibration will be used for the other sensor as well. Then consider this: What would a 'proportional' strategy look like for two light sensors?

Copyright © 2007-2011 by Dave Parker. All rights reserved.
All project designs, images, and programs are protected by copyright. Please see the usage policy.

Fun Projects for your LEGO^® MINDSTORMS^® NXT!

Connect the Color Sensor to port 3 on the NXT.

There are two programs provided for the Line Follower. Both programs use the Color Sensor in the 'Light Sensor' mode, where the sensor shines an LED lamp of one color (red is used here) and measures the amount of light that is reflected back into the sensor, as a number from 0 to 100. Although under good conditions it is possible to follow a line using the Color Sensor in Color Sensor mode, the Light Sensor mode is better for this task.

1. The LineFollow program is a simple 'Two State' line following strategy without any sensor calibration. It will repeatedly turn either left or right in a 'zig-zag' pattern, alternating back and forth on either side of the right edge of the line. Start the robot near the right edge of the line.

To use this program, you will usually want to first determine a good brightness value to use as the 'cut-off' between the two states (light and dark). To do this, you can use the LightMeter program to measure the brightness over the center of the line (darkest), then over the open floor (brightest), then average them to find the cut-off. For example, if the Light Meter reads 20 over the line and 40 over the floor, then 30 would be a good cut-off value.

2. The SmoothFollow program is a much more complex 'Proportional Line Follower' which continually adjusts the steering response of the robot depending on how far off the line the robot appears to be. If the robot is only slightly off of the edge of the line, then only a small amount of steering is applied. When the robot is exactly over the right edge of the line, the robot will drive straight ahead at the full power specified. This allows the driving to be smoother and faster than the LineFollow program.

In addition, the SmoothFollow program does an automatic 'calibration' sequence at the beginning, where the robot scans the line and the surface to determine darkest and brightest light values to expect, so there is no need to manually determine a cut-off value. To get a proper calibration, start the robot with the sensor directly over the center of the line.

• Try designing your own course for the Line Follower, with a mixture of gradual turns and sharper turns. Electrical tape works well because it is dark, flexible, and leaves no residue when removed. Adjust the Power levels used in the programs as necessary to allow the robot to still follow the course. The tighter the turns that the course contains, the slower the robot will have to drive to avoid missing the turns.

• Add the Color Sensor to another robot and see if you can make it follow lines too. Which robots follow lines better than others? Among other things, it will depend on how the wheels are spaced and where the sensor is placed on the robot. Try modifying the 'Gain' variable in the SmoothFollow program to try to tune the steering response needed for your robot.

Copyright © 2007-2011 by Dave Parker. All rights reserved.
All project designs, images, and programs are protected by copyright. Please see the usage policy.