Bamboo is a prototype robot panda built around the Intel® Joule™ with Windows 10 IoT Core.

Bamboo has the following abilities:

• Moves via closed loop control and easy to access busses – GPIO and PWM using off the shelf motor controllers, motors, encoders and libraries.
• Sees and processes the world using depth information from an Intel® RealSense™ camera over USB 3.0.
• Listens via command and control. While Bamboo understands English, you can speak to her in any language, which is translated via Microsoft Translation services into her native English.
• Speaks via attached off-the-shelf USB speakers.

NOTE: Running the code in this repository on a device with Windows 10 IoT Core requires an Intel driver that is not yet publicly available. This only affects the camera and mapping portions of the demo. This notice will be removed as soon as that driver is released.

This repository contains references to submodules. Be sure to use the --recursive flag when cloning.

git clone --recursive https://github.com/ms-iot/bamboo-demo.git

The code is comprised of several projects which are explained in more detail below.

• Bamboo
• Motor
• PidController
• PwmPCA9685
• UniversalMediaEngine

If you wish to recreate this project or even just run certain pieces of the code you will need to configure several services and provide some information about the mechanical parts you have chosen to use if different from those listed in the part list.

You will need to setup a subscription with Microsoft Translator for the multilingual command and control feature of Bamboo to work. Click here to get started.

Speech Translation API Documentation

Once you have set up the translation account you need to enter your client ID and secret token into Secrets.cs.

// ID and Secret for translation service
public const string AzureDataMarketClientId = "";
public const string AzureDataMarketClientSecret = "";

All of the configuration constants for Bamboo are located in Configuration.cs. If you build your own platform you will need to change the values in this file for things like wheel diameter, axle length and motor gearbox ratio among other things.

MORE INFO COMING SOON HERE

NOTE: Because of how the AudioGraph API works the speech and translation functionality will not initialize unless an audio output device is connected to the Intel® Joule™

While Bamboo works great without an attached screen you may want to see the camera view or look at the 3D map that is being generated. If you have access to the HDMI port on the Intel® Joule™ you can connect a display there but another option is to use the Windows IoT Remote Client which allows you to have complete remote control of Bamboo's UI from any Windows 10 device.

This project contains the main UI for Bamboo as well as things like the reminder manager, odometry, drivetrain code and camera depth processing logic.

This class coordinates the actions of both motors to create directional movement like forward, reverse, and turning. It uses PID control to smoothly drive each motor to specific RPM values.

Drivetrain drivetrain = new Drivetrain();

Note: Don't forget to set the drivetrain-related constants in Configuration.cs.

The LED strip used in this project is NOT a NeoPixel device. It is an RGB LED Weatherproof Strip from Adafruit and relies on shift registers to individually address each LED. We created a class to simplify updating each LED on the strip or all of them at once. Because the LEDs are updated via cascading shift registers the entire strip must be updated at once. So while it is possible to change each LED individually an entire strip refresh is needed and is accomplished by calling the Refresh() method after making one or more set calls.

LEDStrip ledStrip = new LEDStrip(int dataPin, int clockPin, int numLEDs);

• dataPin - The pin to which the strip data line is connected
• clockPin - The pin to which the strip clock line is connected
• numLEDs - The number of LEDs on the strip

This class manages the servos and drivetrain. It uses the Drivetrain class to interact with the physical platform.

MovementManager manager = new MovementManager();

The SLAM algorithm in this project requires accurate position and angle information from the drivetrain. Because Bamboo doesn't have an IMU or GPS unit we used a common dead reckoning algorithm to determine coordinate position and angle from only the encoder output. This singleton class could be used in any robotic project that has motors with rotary encoder output. Note: Don't forget to set the drivetrain-related constants in Configuration.cs. Otherwise this class will not generate accurate odometry.

The Odometer is a global singleton that is accessed directly as Odometer.Instance.

This class handles Bamboo's voice command interface. You can change Bamboo's name by simply changing the ROBOT_NAME constant in Configuration.cs.

// Listen for "Bamboo Forward"
SpeechManager manager = new SpeechManager();
await manager.Initialize();
manager.AddCommand("Forward", async () =>
{
	// add logic here
});

SpeechManager manager = new SpeechManager();

This project contains the logic for getting the RPM of the motor shaft and controlling the motor throttle. Throttle is controlled using a PWM-controlled motor controller which is attached to an Adafruit PWM Servo Hat. The RPM value is calculated using an optical encoder and the high speed interrupt capabilities of Windows 10 IoT Core. Knowing the pulses per revolution value of the encoder along with the elapsed time between interrupts we can easily calculate the RPM. We used an exponential moving average to smooth out the noise.

Motor motor = new Motor(0, 36, 250);
await motor.Initialize();
motor.Throttle = 50.0; // Set the throttle to 50%
var rpm = motor.RPM;   // Get the current RPM

Motor motor = new Motor(int motorPin, int encoderPin, int encoderPPR);

• motorPin - The PWM channel on the HAT to which the motor is connected
• encoderPin - The pin to which the optical encoder channel is connected
• encoderPPR - The pulses per revolution value of the encoder used (found in the datasheet)

This project contains the PID logic for achieving closed loop control of the motor RPM and is documented here. The PID loop needs to be tuned to achieve smooth motor control. The gain constants are in Configuration.cs and may need to be modified given your motor and gearbox choices.

// The PID gain constants were derived from trial and error tuning
public const float PROPORTIONAL_GAIN = 0.1f;
public const float INTEGRAL_GAIN = 0.15f;
public const float DERIVATIVE_GAIN = 0f;

This project was copied from the BusProviders repository and it allows us to use the Adafruit PWM Servo HAT to communicate with the Victor SP motor controllers. The controllers require a PWM input signal which is delivered by the Servo HAT. The HAT is connected to the Intel® Joule™ via I2C and creates a nice abstraction for communicating with multiple motor controllers without having to drive the PWM signal directly.

This is a library that simplifies the playing of media in Windows 10 IoT Core and is documented here.

===

This project has adopted the Microsoft Open Source Code of Conduct. For more information see the Code of Conduct FAQ or contact [email protected] with any additional questions or comments.