-
Notifications
You must be signed in to change notification settings - Fork 45
FAQ
- How does the cruise control work?
- How can I optimize the PID algorithm of the cruise control?
- How do I calibrate the gyroscope?
- How do I calibrate the speed encoders?
- How do I use an existing class with a less complicated name?
- How can I add a new sensor in the library?
- Where can I buy an already assembled shield and car?
- Why does the <X> sensor not work?
- How can I use this library with a servo motor and an ESC?
In cruise control mode, the vehicle attempts to move at a stable speed, regardless of various external disturbances. External disturbances could include the floor's friction, the weight of the car, the voltage battery drops etc. It is essential, if the user wants to make sure that the car will move at a specific speed, regardless of those factors. The algorithm utilized in order to achieve that is called PID controller. Nevertheless, the fact that the car attempts to maintain its speed, does not necessarily mean that it actually accomplishes that and particularly in an efficient manner.
For maximum performance, a user should optimize their PID controller, using the Kp, Ki and Kd variables as arguments in the enableCruiseControl method, as well as the frequency that the controller will be accessed.
To put it very simply, you optimize or tune a PID controller by changing the Kp, Ki and Kd values and observing its behavior/output. If the oscillations in the output in its effort to reach the desired outcome are nominal and if the controller reaches the desired state fast enough, then you have a well tuned PID controller.
There are many ways to do that, but for reference you can have a look at the Ziegler-Nichols method. Alternatively, you can have look at some more empirical and practical approaches here: What are good strategies for tuning PID loops?
The gyroscope, even at its idle state (when it is being still), produces an output, which usually varies between a range. This noise can affect the stability of our readings, if not taken into consideration during the calculations from which we derive the angular displacement. Therefore, we need to define an offset (as well as a threshold) in order to filter out this invalid measurements. As you should not expect the offset value to be valid for your own gyroscope, you should run the sketch found under Examples → sensors → gyroscope → gyroscopeCallibration, while having the vehicle still and the serial monitor open. If ran successfully, you will get a value that represents the offset for your specific gyroscope and you can use it from that point and onward in the gyroscope's constructor.
The speed encoders measure how much a wheel or motor has rotated, by counting pulses generated by the infrared beam being blocked and released, due to the movement of the encoder disk. Read more on how they work.
In order to derive the traveled distance from these pulses, we need to know how many pulses we get per meter or centimeters. We can use that number as a factor in order to calculate the distance covered by the vehicle. For example, if we know that we get 50 pulses per meter, then we need to multiply the amount of pulses that we get by 2 (alternatively, divide that number by 0.5), in order to estimate the distance we have traveled, in centimeters.
As this number depends on the encoder disk, as well as the sensitivity of the odometer sensor itself, you should not rely on the default value, provided with the library. Instead, you should discover that number for your own setup. You can do this by uploading the sketch found in Examples → sensors → Odometer → findPulsesPerMeter. Next, take a measure of 1 meter and place it on the floor. Having the serial port open, either by having the car connected to your computer with a preferably long USB cable or using Bluetooth, roll the car along the measure and stop it after its wheels have traveled (approximately) one meter. Do not forget to make sure the wheels are spinning and not sliding on the floor. You might need to apply some light pressure downwards, in order to manage that. The number that is printed on your serial monitor, is the one you are looking for that represents how many pulses we get per meter from the speed encoders. To be safe, do this a couple of times, so to be certain that the number you have found is the correct one.
Finally, you should provide that number as an argument in the Odometer constructor, so you can get accurate distances in centimeters and speed in meters per second, while using the appropriate functions.