A couple of years ago I first saw one of those self contained game systems that consist of just the joystick and the AV wires that you plug into your TV. I was simply amazed at the simplicity and brilliance of the concept. You just plug it in and play.
Nuntius is based on the same basic concept of those systems except it’s a robot. The controller it a biomechanical input device that consists of a propeller proto board that measures the position of a 5 axis +gripper “mini arm” that you articulate with your hand just like you would if you were actually there. The propeller then send the data via a XBEE to the robot. The result is that the big arm on the robot mirrors the mini arm and you control the robot by looking through a wireless camera that’s attached to the gripper. The movement of the robot is proportional to the direction and how far the arm is extended (or pulled back for reverse). The further the arm is extended, the faster it goes in that direction. The robot only moves while you’re holding down a button on the proto board. This allows you to drive the robot around and then stop and examine something with just a button and the mini arm. You just plug it in and do whatever… weed the lawn, weed the garden, weed the driveway, take the dog for a walk… from the safety and comfort of your couch.
Dave
Major components:
2 Propeller proto boards
1 Propeller proto board enclosure
2 XBees
1 Parallax Wheel Kit
2 Parallax HB-25 Motor Controllers
Servos and Servo Brackets from Lyxmotion
Wireless camera
David, where have you been all my life? I have a G-scale garden railroad occupying a space 130’ by 8~20’. I have spent the last several days feebly attempting to clear the weeds. So now you show up!
Could you make a few minor changes please? I would like it to be completely autonomous, so I don’t have to wear out my wrist. Then could you add some weed recognition software coupled with desireable plant recognition software (sort of a horticultural IFF Identification Friend or Foe system) so it doesn’t shoot down expensive dwarf plants and ground covers. Oh, it also needs to climb over and into spaces bounded by 2~4’ walls. And to protect my track and buildings it should follow Asimov “First do no harm.” Or was that Hippocrates? A sensor for ground moisture would be a nice plus too.
I host my train club in three weeks. If I could have it by then, you’ve already done the hard inventve work, I might interest a few more takers.
I loved the humor of your video, especially the startled first sighting of the weed and then the build sequence.
I’m glad you liked it and ya, that should only take me a couple more days for those “small” features (although I have thought about making an autonomous weeding robot for commercial row crops like corn or soybeans. I think the key to making a weeding robot isn’t so much one that can identify weeds [there are far too many variants], but a robot that can identify the crop. This would be relatively easy for corn and could probably get 80-90% of weeds in a normal garden.)
I’m glad you thought it was funny too. Did you see the blooper at the end? A friend of mine gave that to me as a gag gift for my birthday at about the same time we had discovered “Conan the Musical” on YouTube.
The idea of the controller was what made me interested enough to do the robot. The controller’s mini arm is made from 6 10K ohm linear resisters that are part of a capacitor/counter circuit. The circuit works because in each of the propeller’s eight identical processors called “cogs” are two counters that can be set-up to work in the background of each cog without the bothering anything. They can produce VGA singles for a TV or AD/DA conversion. In this case each 10K pot is connected to it’s own pin on the chip along with a 0.01uF capacitor and in parallel. The code works by first setting the pin to output for 1/100,000 sec which charges the capacitor. The code then turns the pin to input and starts the counter. The counter is set-up to return the value of the “counter” once the capacitor discharges (through the variable 10K pot) enough that the voltage drops from high to low. The greater the resistance, the longer it takes to discharge and the larger the “count” variable will be.
Anyhow, I hope your train club appreciates your “all natural” weeded yard . Have fun!
I also loved the video. This is a very cool project!
We are actually working on a kit that includes the teach pendant like part. We are using modified servos for the pots. The idea is to be able to use the same brackets to build the controller arm as we use in the arm itself. You can even use the motor and geartrain to add tension for the axis. Shorting the motor leads provides the most tension. Leaving them open provides very little. You can add resistors to fine tune the tension. Again, nice work!
My first thoughts were to use the SES for the mini arm, but I wanted to make it fairly small. I also don’t know how to do what you’re doing with the servos, but it sounds very interesting and I would love to hear/see some more. My only advice with your own project is to make the control arm about the same size as the real arm. Having the control arm so much smaller than the robot makes fine movements difficult.
Thanks again for making great quality stuff to buy. It lets me build more with the time I have.
(although I have thought about making an autonomous weeding robot for commercial row crops like corn or soybeans. I think the key to making a weeding robot isn’t so much one that can identify weeds [there are far too many variants], but a robot that can identify the crop. This would be relatively easy for corn and could probably get 80-90% of weeds in a normal garden.)