to mark a step as The goal of my project is to create a robot pet that works indoors.(Not a pet robot!)
I think that pet robots are robots similar to animals (such as dogs and cats).
However, what I'm trying to make is a pet as a robot.
The following functions can be thought of as necessary functions to achieve this goal.
- Moving mechanism(Mapping,Autonomous traveling etc.)
- Cute look
- Recognition function (audio, camera etc.)
- Reaction function (movement, voice etc.）
I am developing to implement these functions.
I keep the development record on Youtube(https://www.youtube.com/channel/UCgEETDZfbyr0_9rYclqnPsw).
System configuration(I can not upload images for some reason so I will link the image)
I adopted the Mecanum wheel. Because it becomes the same as the pet robot in the walking mechanism.
So, I will use a Mecanum wheel that can move in all directions like a robot．
Mecanum wheel
https://www.robotshop.com/en/60mm-mecanum-wheel-set-2x-left-2x-right.html
I controlled by referring to the following page.
https://seeeddoc.github.io/4WD_Mecanum_Wheel_Robot_Kit_Series/
I used Aruduino pro mini and TB6612FNG for control robot.
Arduino programs:https://drive.google.com/open?id=1puwvaQ0M6EIU0JviEIyd0kBxR8JDYrvE
At first I was using TA7291P for MD, but I changed to TB6612FNG because of the large loss.
Power-supply voltage：5.08 V
PWM frequency(Output from Arduino pro mini) : 31.25 kHz
TA7291P Output voltage :
3.90 V(100%), 3.16 V(80%), 2.43 V(60%), 1.69 V(40%), 0.95 V(20%)
TB6612FNG Output voltage :
4.44 V(100%), 3.71 V(80%), 2.96 V(60%), 2.21 V(40%), 1.49 V(20%)
On average, TA7291P has a lower output voltage of 0.54 V than TB6612FNG.
I am planning to compensate using IMU to go straight.
I used Madgwick filter to calculate Quaternion.
http://x-io.co.uk/open-source-imu-and-ahrs-algorithms/
Then, in order to autonomously travel inside the room, Raspberry pi 3B + (Raspbian Stretch + ROS kinetic) was used for controlling the entire robot. Furthermore, in order to use the autonomous running package, visualize the state, I prepare a PC(Virtualbox(Ubuntu 16.06 + ROS kinetic)) outside and communicate via ROS.
This video is a state of autonomous traveling using parameters before tuning. I used gmapping to create the map.
At this stage I am doing with non-holonomic behavior.
The next video is autonomous traveling after tuning parameters. The map used is the same.
It is done with holonomic behavior.
move_base parameters is here:https://github.com/ranbo0311/Mecanum_robot.git
ROS configuration
Although there are still incomplete parts, the movement mechanism is almost complete.
The next step is recognition.
This step is still on its way.
First , color tracking was done without ROS.
Color tracking was done using picamera v2 and OpenCV.
I am using the controller to operate the robot being tracked in the video.
What I am doing now is simple color recognition, but I would like to do face recognition, object recognition, etc. in the future as well.
Next, AR marker tracking was done using ROS.
I used OpenCV aruco in Python to detect AR markers.
It is expected to use a USB microphone and julius.