🧵 View Thread
🧵 Thread (33 tweets)
I hope they release their code. Planning for my own vision loss and the cost of a dog is a huge problem. I can make a DIY robot like that for less than $2000 and a seeing-eye dog is $50k.Slam based system they're using also would help with wheelchair. https://t.co/h1zd9hHgaI https://t.co/WJWqd6Duin
I could probably train the seeing eye dog myself, but the whole occasionally not being able to get out of bed or walk puts a dampener on my ability to take care of a pet. https://t.co/X9MrFLJixa
Literally want a robot that can help me navigate autonomously. https://t.co/oYxyNH091g
The next thing I'm doing? Using those PVC skills to build myself a wheelchair that fits that comfy 'baby' car seat for when I eventually succumb to an M.s. flair-up outside of my ability to stop.I'm thinking of something just like these kids's seats:https://t.co/31BlBgembY
If you aren't in academia, but have some microprocessor skills, want to play around with this, pick up a used roomba and play around with one of these. They're only $99 at adafruit.https://t.co/deWB0NEJALhttps://t.co/IE5a4WV2kp
@Smingleigh @KyleBrubaker a new edition to the RPLIDAR line was just released. It seems that means the older models are getting cheaper. Adafruit has the bare bones ones for ~$115 (out of stock), and there might be some hitting ebay as people upgrade. Sparkfun has the other 2.https://t.co/AQWmI7B1wd
The larger problem that needs to be solved is localization and navigation. The actual hardware is trivial.DroneBot workshop has a good intro to LiDAR.https://t.co/kwdIsiNqR8
If you want a cheap inertial sensor and gps, a used phone combined with a lidar device can would get you pretty far. The keywords to search for are "sensor fusion" https://t.co/G0hruUBVSo
The nice thing about mounting this to a robot dog is that a 4 legged robot is much better able to traverse a typical human living area. https://t.co/a4OK4bOdhF
Lidar alone isn't really good enough to give you full navigation or obstacle avoidance though.This is a good review of limitations of the simple lidar. https://t.co/P6QLGCOwD2
The benefit is that it's cheaper than a full vslam system while still getting you enough detail to navigate a human scale room.https://t.co/kgVSb2FV8J
Don't worry, I'm sure these designs won't be used to hunt people down and kill them. I hope.https://t.co/L14yRUG6Ta
That video is about a mapping and tracking **prototype** someone built for a thesis exploring a human-computer collaboration. Of course if you wanted a production one you'd use something like Ghost Robotic's Spirit or better fit for outdoor use.
What were they thinking not including a real blind person when testing if a head tracking camera can be used to correlate a human's position in space and orchestrate a guided walk as a proof of concept!The nerve of them for doing basic robotics research for a thesis...
Seeing people comment on the video about how this will never work for a real world use case - focusing on the robot.Someone asked if the robot was waterproof... A prototype built on a off the shelf dog frame to test a bit of of software?!
Are people acting in bad faith? Or is the entire fucking world going mad?Come across a robot prototype that guides people thru a room, and now we have people complaining that the thing can't handle getting wet and we haven't tested it with blind people?
I am however, acting in bad faith. What you really need to be paying attention to is how these small projects are acting as stepping stones for fully autonomous kill drones.Look at this one! they made it jump over obstacles.https://t.co/5dchyQ1g6Z
So figuring out how to use a robot to guide a human is actually a complex problem. They spent a lot of time figuring out how to get the robot to provides lack on the line while it explores the environment, thus avoiding the human walking into things.https://t.co/jDLq0u5Nnz
You only have to solve this once for a robot. For a guide dog, you have to train the dog not to pull the owner off balance for every single dog you train.But maybe we could just automate away the training of dogs? I have also explored, and want that.https://t.co/2UJnTXt334
I have personally trained a dog to help my mom walk.They are called "mobility assistance dogs" and if I could get a diagnosis for my M.S. I'd probably get and train one for myself.https://t.co/Yh6TSSlY30
Sometime in the next 5 years we're going to get someone complaining the ADA only covers dogs as service animals. We'll probably have to reclassify robotic dogs as medical devices. Think 70k for a robot dog is $$$ now... Wait till you see effects of needing to jump thru FDA regs.
I want a tiny death machine to help me walk.https://t.co/k64UrO3FqE
2011 was such a hopeful time for me.https://t.co/DMQcUULS2F
Even simple LIDAR is surprisingly versatile.https://t.co/scflWUduVg
Cute little mobile death machine.https://t.co/aolktDsDEt
hexcopter + robot cooperation.https://t.co/cD5WgMH9Sa https://t.co/zqWERd7zmh
While I am able to walk better right now due to my medical experiments seeming to work out. I still want a wheelchair, just in case.https://t.co/ptruYPYpHI
I want to build things again.https://t.co/xnJgEweqkm
LiDAR is going to be a whole other level once we can use this for sub micron precision and rapidly bounce different frequencies of lasers off walls to see around corners.https://t.co/DXNXMN6lzU
"Navigating Real-World Challenges: A Quadruped Robot Guiding System for Visually Impaired People in Diverse Environments" [2024]https://t.co/WCpTdjyUPn
"Take me to the Chinese food store"https://t.co/kIh3hzFFvr