Added telemetry to the toy car

This is a followup of my original post.

I have added a voltage readout of the battery in my toy car. It looks like this on client applications:

Java Swing application

Android application

Voltage is read using MCP 3008 A/D converter:

MCP 3008 A/D converter pinout

Here is the layout:

• connect pins 16 (Vdd) and 15 (Vref) to the 3.3V pin of your GPIO port of your Orange Pi, or Raspberry Pi (they are pin-to-pin compatible),
• connect pins 14 (Agnd) and 9 (Dgnd) to the GND pin of your GPIO port,
• connect pin 13 (CLK) to the pin 23 of the GPIO port,
• connect pin 12 (Dout) to the pin 21 of the GPIO port,
• connect pin 11 (Din) to the pin 19 of the GPIO port,
• connect pin 10 (CS) to the pin 24 of the GPIO port.

Make sure that SPI is enabled on your device. On my Orange Pi, it is was enabled by default. On Raspberry Pi, you need to start the raspi-config and enable SPI from the menu.

If the SPI is properly enabled, you will be able to see two files in the /dev folder of your Pi:

/dev/spidev0.0 and /dev/spidev1.0.

I have used Adafruit MCP library from this location.

When you download and install this library, you will use it by importing following modules:

import Adafruit_GPIO.SPI as SPI

import Adafruit_MCP3008 

Then you need to set it up:

# Hardware SPI configuration:

SPI_PORT   = 1

SPI_DEVICE = 0

mcp = Adafruit_MCP3008.MCP3008(spi=SPI.SpiDev(SPI_PORT, SPI_DEVICE)) 

On my Raspberry Pi, I have set the SPI_PORT variable to zero (0). On my Orange Pi, I had to set it to one (1). You can then read the voltage this way:

voltage = mcp.read_adc(0) 

The zero argument from the read_adc function is the number of the channel. In my case, I have connected the voltage from the battery to the pin 1 of the MCP3008, which is channel 0. Since the voltage of the fully charged battery can exceed 10V, I had to reduce the voltage:

Voltage reduction

I have extended my original Python server for the toy car to accept a telemetry request (string 't' as a telemetry request) and to send the voltage (currently it is just the voltage), as a JSON string:

elif text == 't':

voltage = mcp.read_adc(0)

c.send("{ \"voltage\": " + str(voltage) + '}\n')  

My client applications (both Java Swing and Android) send periodically this request string and receive a JSON string. Then they parse the JSON string, extract the voltage, and display it on the screen.

Orange PI Zero and the Toy Car

This is a followup of my original post.

As I promised in my previous post, I am writing this post to share my experience with the Orange PI Zero and my toy car. The board is very small, and has almost all you need, except for the HDMI video output, which I don't need for my project. Therefore, it is perfect for me.

First of all, I have placed a small heat sink ond the SoC, since it can get quite hot (sometimes around 65°C). Next, I have placed the board on the car chassis and secured it with two screws. OPI Zero has both Ethernet and WiFi onboard, and even has the built-in antenna.

Since it does not have the HDMI, the only way I could configure it was to use the serial port. OPI has excellent support for serial console. Next to the Ethernet connector, there are three pins for Tx, Rx and GND. Thanks to that, I was able to set my WiFi using serial console, and from that moment on, I could SSH to it via WiFi.

OPI board 

The car now looks like this:

With everything added...

GPIO port does not have pins, so you need to solder them yourself. I have soldered seven pins (4 for the motor control, 2 for +5V/GND, and one for the signal which turns on/off headlights). You can see those seven pins on the top right corner of the picture above.

Next came the software. I want to state that the armbian support for the Orange PI is perfect. Everything works out of box. When I have configured the WiFi to connect to my home router, I was able to install all the software via SSH.

First I had to install the GPIO support. I have dowloaded and installed the orangepi_PC_gpio_pyH3 library, made by the duxingkei chow:

https://github.com/duxingkei33/orangepi_PC_gpio_pyH3

This library is similar to the one I have on the Raspberry Pi. Not the same - just similar enough.

The next step was to install the mjpeg streamer. I have installed it, but it could not run, saying that the libjpeg is missing. The only way I could make it work was to manually download the libjpeg8_8d1-2_armhf.deb file and to install it from that .deb file.

Next came the new kind of problem: mjpeg streamer crashed with this misleading stupid error message, when I tried to stream from two cameras at the same time:

Unable to start capture: No space left on device...

This does not have anything with the free drive space. It simply says that the complete bandwidth of the USB controller is consumed by the first web cam and it cannot stream from the second web cam. It simply cannot work with two Logitech C170 cameras.

I have tried to set the number of quirks for the driver, but it didn't work.

Fortunately, I had one Logitech C210 web cam (lower resolution - lower bandwidth) and that was good enough for the poor USB controller on the OPI Zero: one C170 (forward cam) and one C210 (rear cam).

The WiFi antenna is not so powerful as I have hoped, but it indeed works better than the small Realtek USB dongle which I had previously. The Ralink-based dongle and antenna I have described on my first post works the best, but I already have the built-in adapter and the antenna, so I will stick to that one.

That is about it. I am very satisfied with the Orange PI. I have tried both Lite and Zero boards, and they work excellent, considering the price and features. The only drawback of the Zero model is its limited USB bandwidth and the high core temperature. But, I can live with it...

Toy Car v2.0

New version of the RPI-driven car

This is a followup of my previous posts: this and this.

I wasn't satisified with my previous build of the car. The bunch of wires was sticking all around the car. More important, the car just had a chassis with everything thrown on it. I wanted to have a complete car (with panels, headlights, etc.). That idea required to cut some wires, to solder instead of connect, to glue those wires with the plastic gun and so on. 

The result is here:

Here are some snapshots:

It looks nice, doesn't it?

I managed to put everything inside the car. Cameras, too.

 The front camera

 The rear camera

I started with the chassis. But this time, I have cut the wires to the appropriate length, used shrinking plastic insulator and screwed some components on the chassis:

The car inside looks neat now

I have used existing switch (from the original car setup - bottom left corner) to turn on/off the car. The switch is connected to the battery connector, so it controls the whole car. Now, when I want to use the car, I have just one connector to connect to the battery (top right corner of the photo).

The final assembly looks like this:

Added USB cams, headlights, A/D converter and serial port

I have ordered Orange Pi Zero to put instead of Raspberry Pi. I will write my experience with it when it arrives.

Raspberry Pi drives a toy car

Playing with Raspberry Pi

Followups:

- added telemetry to the car

- Orange Pi Zero and a car

- complete rework

- added headlights to the car

I have used the Raspberry Pi in this project to control the toy car. I have a long history of remotely controlling stuff. Almost thirty years ago, I have made a board with relays to remotely control the toy car using ZX Spectrum 48 computer. At that time, I have purchased Z80 PIO chip, made a PCB board and soldered transistors, resistors and relays so I could drive the wired-controled toy car.

Last couple of years I was thinking of doing the same, but with the wirelessly-controlled car. I didn't want to use existing electronics and remote control that came with the toy car. I wanted to control the car using my computer.

I was looking at the old remotely controlled car which my kids used to play with. All the electronics was either broke, or missing. I just had the chassis with two motors: front to steer the car left/right and the back one to move the car forward/back. I figured out that I could use that car to play with. All I had to do was to get some controller and to wire that controller to some small computer that could be placed on the car. I was thinking of putting the smart phone on the car, but then the Rapspberry Pi came. Pi is much better choice, since it already has GPIO ports, while smart phones need some extra hardware (for example, the Android has IOIO hardware for that purpose).

So, this was my initial plan: get the Pi, get the controller, mount both on the car and make some software which would allow me to drive the car using my PC, or smart phone, by sending commands via WiFi. If I could mount some USB web cams on the car, even better.

This is the initial result:

The first car

I have purchased the controller based on the L298HN chip and placed that on the chassis:

It is quite easy to control the car using this piece of electronics. There are four pins at the lower right corner and they are used to control two motors: first two pins control the left/right motor, while next two pins control the forward/back motor.

The second car

I have purchased a Raspberry Pi to control the car. When it arrived, I have placed it on the car, but the car could not drive on carpets because the wheels were too small. So I bought a new toy car with bigger wheels and stripped all the electronics:

This was the initial configuration:

There is a LiFePo4 battery, voltage converter to 5V, with two female USB connectors, a Pi, and a driver. The battery has three cells, and gives 9.9V. I have put three diodes in series to lower the voltage for the motors (quick&dirty solution).

I have plugged the USB WiFi dongle into the Pi, so could I connect it to my home network. That way, I could send commands from my PC or smartphone and therefore control both motors (drive the car).

The first thing I did was to copy the Python test program to the Pi to see if I could control motors using the Python code:

#!/usr/bin/env python

import RPi.GPIO as GPIO
import time 

PORT_BACK = 26
PORT_FORWARD = 19
PORT_LEFT = 13
PORT_RIGHT = 6

GPIO.setmode(GPIO.BCM)
GPIO.setup(PORT_FORWARD, GPIO.OUT)
GPIO.setup(PORT_BACK, GPIO.OUT)
GPIO.setup(PORT_LEFT, GPIO.OUT)
GPIO.setup(PORT_RIGHT, GPIO.OUT) 

GPIO.output(PORT_FORWARD, 1)
time.sleep(1)
GPIO.output(PORT_FORWARD, 0)

GPIO.output(PORT_BACK, 1)
time.sleep(1)
GPIO.output(PORT_BACK, 0)

GPIO.output(PORT_LEFT, 1)
time.sleep(1)
GPIO.output(PORT_LEFT, 0)

GPIO.output(PORT_RIGHT, 1)
time.sleep(1)
GPIO.output(PORT_RIGHT, 0)
GPIO.cleanup() 

As you can see, it is quite easy to control motors using Python code on Pi. All you have to do is to send 1 to the port to turn on the motor, and 0 to turn it off.

The next thing was to make a Python application that would be a TCP/IP server. It would wait clients to connect and then would wait for commands, in form of strings. Clients would be either PCs, or smartphones. I have created a simple protocol to control the car. It is text-based and clients just send short strings which Python server receives, parses and executes. Here is the list:

• F - turn on the forward/reverse motor to go Forward,
• B - turn on the forward/reverse motor to go Back,
• x - turn off the forward/reverse motor,
• L - turn on the left/right motor to go Left,
• R - turn on the lett/right motor to go Right,
• X - turn off the left/right motor,
• exit - client disconnects,
• shutdown - shutdown the Pi,
• reboot - reboot the Pi.

The client application can send combinations like: 'FL', or 'BL', meaning 'move forward and left', or 'move backwards and left'.

Then I have added two web cams: one for the front view, and the other one for the rear view:

I have installed the mjpegstreamer application which streams live webcam videos from both cams on the web server (two ports:8080 and 8081).

Then I have upgraded my client applications (both Android and Java Swing) to work with live streaming coming form the Pi. Both applications can connect to the Python server and mjpegstreamer at the same time. They show the live video stream from both cams, and send commands from the keyboard, or gamepad to the Python server:

Android app

Java Swing app

Schematics

Here is a simplified schematics:

Very simplified schematics

WiFi dongle problem

The USB WiFi dongle which people usually buy for the Pi has very small antenna (Realtek RTL8188CUS chipset). That antena is so small, that my car used to lose the WiFi connection in rooms not close to the home WiFi access point. That is why I have purchased another WiFi dongle, but with the bigger antena on it. However, this one had the Ralink chipsed, instead of Realtek. That was the problem since there was no support for the Ralink-based WiFi dongles in the Linux kernel prior to v4.0 (all this happened last year). So I had to make&install Linux kernel driver for it... Not the fun stuff. When the support for the Ralink dongles finally arrived (with the 4.0 kernel), I was able to use the dongle without building kernel drivers. Currently I have a Linux kernel v4.0 on my Pi, and the dongle works out of the box.

USB and Power problem

On my Pi I had three USB ports filled with devices (one WiFi dongle and two webcams). Pi cannot provide enough power on its four USB ports for all the devices I plugged in, so I had to put the powered USB hub. I had a spare female USB connector from the 5V voltage converter (the first one was used to power up the Pi). I have connected the power input of the USB hub to that other female USB connector:

The powered hub is below the webcam (the blue box). All USB devices are now plugged in it.

The current configuration looks like this: