Build comfortable(ish) battery operated exoskeleton or support harness that can help my dog walk a bit more stabler and help with the bathroom breaks and stumbling.
I have a very athletic and a bit nuts dog that somehow blew her ACL. We did the surgery but she will never run again let alone go to bathroom which is still a bit of a hassle and requires me to basically carry her several times a day, and we're talking a 75lb dog here.
I've come up with all sorts of potential shoulder straps and wheelchair contraptions based on those bike trailers but in the end the exoskeleton is a harder project but combines a bit of my robotics knowledge, with learning some new things since I already have a ton of parts and 3D printers laying around. In the end it will hopefully help both my dog and I to get a breather and a better quality life.
Measurements and First thoughts:
I've taken some measurements of the back leg, hip, stomach, around the belly, from neck to butt. etc. basically every measurement i could do manually but I'm aiming to take the structure or kinect out and actually mapping her if she'll stand long enough for me to do that..
Speaking of standing she's kind of lopsided so maybe I'll put an actual stand underneath her to hold her up for a few minutes..
When to poop or not to poop:
Next are the harder details. She's a girl and this took me a while to get used to the whole not lift leg on fire hydrant issue but that means she squats to pee. This means that pooping and peeing are similar other then curbing the back. I've seen a few articles on using gyro's and IMUs combined with GPS to track where their dogs poop in the yard so they know where to go pickup.
I've always wanted to do that project so I think incorporating those IMUs on the back left leg (wobbly), the hip and the shoulders at the very least will provide the proper tilt info for peeing, vs pooping. I don't need the GOS since i'm... well tethered to her anyway.
The point is having the data to know and hopefully later to anticipate what she's going to do and how the exoskeleton should adapt. This data will probably also help me figure out how to even structure the exoskeleton at least with stress points etc how much she drops etc..
I already have a few IMUs for arduino/pi and some metaware i kickstarted a few years ago all with accel,gyro etc.. 9dof on all of them i think which is more then enough. I'm gonna try to make a small harness and leg wrap to keep on her all day and start gathering info.
Exoskeleton and help and ideas needed:
I'm probably missing a ton of data and ideas but this why I'm writing here. I honestly haven't seen much on google for diy exoskeletons presumably because it's both expensive and not very useful in daily lives. Let alone a dog one though i've seen and heard a ton about handicapped and wheel chaired dogs so who knows..
I'm a bit lost on the locomotion and the "anticipation" for lack of a better word of her movements. How do we figure out she's about to lift and put that leg down? I'm guessing based on the other legs moving probably with imu's on all of them ?
Electromyography is another option and is relatively cheap but from what I can tell she probably needs to be shaved and i have no idea what dog/vs human muscle function is. Though it seems more reliable to read a muscle twitch in my mind to anticipate her moving the leg up the anticipating it via IMU...... I leave that to people with experience with this device to help me.
While i'm a bit versed on robotics I'm lost here on what to use for support material. Obviously past the padding, what should I use? First thing that came to mind are linear activators at say close to a right angle probably two on the leg is the easiest but that feels too simplistic and maybe jumpy/heavy ? I've also seen these air muscles used as exoskeletons where you inflate/deflate to provide support though no idea where to start there.
Serial servo's could be another example but it's harder to bend with those just to move up and down. I'm guessing she needs support lifting the left and putting weight on it from how she walks.
Brains and OS potentials ?
I have a pi3 which should be more then enough, though I also have a few esp32, esp8266 and chip pro I was using to teach at the maker space. Though I'm a bit rusty my best guess is ROS will provide the proper coordination with laptop and rapid adaptation I need and probably has a ton of useful plugins and projects to simplify this.
Final thoughts and data to come:
I've sort of outlined the planned project idea. After the 2k+ surgery i'm broke so I'll try using what I have but I do know I can't keep holding a 75lb dog everyday. I have severe arthritis on some days and already had a few shots since then in shoulder from the pain so some solution better then none.
I'll post a video of her walking on her own and also post IMU data when i have it to get help combining the data. Structure sensor data as well if I manage it.
What other data points can i provide other then measurements ? Can this community help me come up with a plan to test and implement? I could 3d model it with blender or fusion360 once there's a basic idea in mind though i'd need help with simulation..
Sorry for long message and any help appreciated to come up with a plan.