Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
[-]
Slideshow

[-]
Stats
» Members: 450
» Latest member: mclaren99uk
» Forum threads: 479
» Forum posts: 2,744

Full Statistics

Custom Drives Ideas
#1
I have been running over in my head the custom drive setup I am intending to build for my R2. I thought I would start a thread to document my process.

I really like the look of the Q85 4wd brushless drives that are being used in some of the newer droids over on the main forum. I wanted to come up with a similar design, but hopefully at a lower cost.

I have ordered some brushless motor hub motors from Aliexpress but am worried that I have bought non gear reduced ones, and while an R2 with a top speed of over 30kph might be cool (and a terrible idea), it may impact his maneuverability at low speeds. I will have to see if the motors still have enough torque to move him around on carpet, etc. Hopefully the lighter weight of a 3D printed droid will help here.

Once the motors arrive I will start work on designing a frame that will hold them and drop into Michael's foot shell design.

One of the big advantages of going hub drive is that the battery boxes remain empty and can be used for batteries. I am planning on packing my battery boxes with 18650 cells. I am still tossing up between doing a custom spot welded pack, or making some sort of battery box where the batteries connect with spring contacts. Having the batteries in the battery boxes means they won't conflict with the 2-3-2 mechanism inside the body.

I am wondering if the added complexity of 4wd is needed, I have heard it helps a lot with driving on uneven ground, which can lead to the driven wheel lifting off the ground. Does anyone with a completed droid have an opinion on this?
Reply
#2
My hub drive wheels turned up today. Now I need to start designing a frame to hold them. I hope 3D printed plastic is going to be strong enough, I hear this type of motor can have a lot of torque.

[Image: attachment.php?aid=552]

I need to order a motor controller for them. I was leaning towards using hobby ESCs but with the upcoming upgrade to the Odrive firmware properly supporting this style of motor I think I am going to use Odrive robotics motor controllers, possibly one in each foot to avoid having to run the motor wires all the way up the legs.


Attached Files Thumbnail(s)
   
Reply
#3
Its just a thought, but
if you use an ESC mounted in the foot area, with the 18650 batteries in the box attached to the foot, you then only need the light weight control wires for the ESC running up the legs.
You may also want to consider a slightly heavier wire to be used for charging the battery boxes from the main unit as well.
You will find readily available from ebay sellers, battery cell balances for the purposes of charging the batteries.
I would also suggest adding a position encoder to the wheels as well for better control
Reply
#4
The wheels I have bought already have Hall effect encoders built in, and with the odrive controllers you can just issue velocity commands and they take care of the rest.

I have been looking at BMS circuits from AliExpress which take care of over current, balancing, over and under charge, etc.
Reply
#5
What I was thinking was more along the line of making sure both left and right wheels turn the same amount and correct for and errors that may occur as a result of differing surfaces as the droid is moving.
While a speed controller on its own will correct the speed after a short delay, the course of the droid can be altered.
A controller watching both wheels at the same time can quickly correct to ensure the course is maintained.

While operating under remote control, errors in direction tracking will not be a problem, but if you fully automate the droid, it can become a big problem.
Reply
#6
(05-29-2018, 09:41 AM)Cyber_One Wrote: What I was thinking was more along the line of making sure both left and right wheels turn the same amount and correct for and errors that may occur as a result of differing surfaces as the droid is moving.
While a speed controller on its own will correct the speed after a short delay, the course of the droid can be altered.
A controller watching both wheels at the same time can quickly correct to ensure the course is maintained.

While operating under remote control, errors in direction tracking will not be a problem, but if you fully automate the droid, it can become a big problem.

Ah, I get you.

I don't think I will be trying to automate my droid. I value it too much to let it run itself into walls. Smile
Reply
#7
he he he, yeah, I agree with you on that point, however, i do plan to give my R2 some intelligence and sensors for collision avoidance.

But during the early testing i will have a remote connected to allow my to stop R2 should the droid do something wrong :-)
Reply
#8
One of my workmates has been trying to convince me to automate mine, but I think just getting it working well on remote might be enough of a challenge, plus I don't want to clutter up his classic design with visible sensors.

The design begins. So far looks like it will fit in the foot shell easily.

It should leave plenty of room to also have the motor controllers mounted inside the foot shell, leaving the battery box entirely free for batteries.

[Image: attachment.php?aid=556]


Attached Files Thumbnail(s)
   
Reply
#9
I'm still working on both the dome and the body at the same time, I haven't yet started on the legs.
I plan to build the new retractable center leg, but from a design and control perspective, I think we need to be able to drive to tilt of the foot relative to the legs, all 3 of them.
also i think the side legs need to be able to extend and retract at least 100 - 200mm to allow for stair climbing.
Sensors have now gotten to be very small and can be easily hidden within the skirt of the droid.
I also plan to install a camera in the Radar Eye.
Using a Raspberry Pi 3 or two, the MyRobotLab system ha support for Open Computer Vision (OpenCV) and the development of deep learning software such as You Only Look Once (YOLO) is progressing well.
Using this tech, the droid will be able to identify objects and people.
Speech recognition already exist along with a basic AI known as ProgramAB based on the Alice Bot AIML language.
This is what the Humanoid Robot InMoov is based on.
Have a look at http://inmoov.fr/ and http://myrobotlab.org/
You will see the developments there that can be used to bring R2D2 to life.
Reply
#10
My design for my custom dive system is coming along. I have scaled up Michael's omni wheels to 133% and modified to fit standard bearings.

My first foot shell should finish printing tonight, and hopefully I should be able to print a test drive frame tomorrow.

Design render.
[Image: attachment.php?aid=557]

View from underneath
[Image: attachment.php?aid=558]


Attached Files Thumbnail(s)
       
1
Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)