Print PDF
Favorite
Email
The idea was to invent an Arduino-based, robotic toy that my dog could interact and play with. One that could record scores, automatically deliver treats, and grow more advanced as the dog learns to play with it. I wanted to find a task to test my dog's intelligence, so I chose a simple system with three 'levers.'
At first, as soon as the dog touches one of the levers, treats are delivered. Eventually the dog should begin to associate the action of pushing the levers with the rewarding stimulus, a tasty treat. Depending on your dog and what really gets him going, you can use light storebought treats, chopped up hot dogs, shredded chicken, or you can use this as a way to make mealtime more interesting for dogs with problems eating. At its most advanced, an LED lights up above one of the three levers and the dog must touch the correct lever in order to get a treat. It is designed to get more and more complex during a single playing event, and after every reset, the dog has to start over. A small counter would track his overall high-score of levers in-a-row, if he could ever actually hit two in-a-row...
There are some pretty widely-accepted theories on the intelligence of dogs generally used to classify dogs. The theory identifies three types of intelligence: instinctive intelligence which describes the activities the dog was bred for, adaptive intelligence, a metric for problem-solving, and working intelligence, or the ability to learn from human interaction. I wanted to come up with a robot that could work with all three types of intelligence for some mental stimulation. Plus, tossing treats at Henry was getting a bit boring and we needed to spice it up.
The larger scheme for the idea of robotic dog toys is to have a centralized treat-delivery system that wirelessly communicates with several toys of varying design, so this 'Touch' module with the levers could just be a single toy in a group. Before I get into the specifics, check out this video of my dog Henry, playing with the finished prototype, so you can get an idea of how it works.
This is the current stage of Henry's interaction with the toy, and now I am working on him using the machine without me around.
The last and most important advice is to SUPERVISE YOUR DOG and do not leave them unattended with any of the items in this instructable. It is not meant to be an autonomous robot that wakes up and plays with your dog, but a game with which you both interact. The fun part for me was teaching Henry how to play with it, watching him systematically improve at the game, only to completely forget everything the next day.
The whole thing took about two months to complete and in the meantime I was able to spend some quality time with my dog and learn that he was pretty quick to pick up how to work the toy!
At first, as soon as the dog touches one of the levers, treats are delivered. Eventually the dog should begin to associate the action of pushing the levers with the rewarding stimulus, a tasty treat. Depending on your dog and what really gets him going, you can use light storebought treats, chopped up hot dogs, shredded chicken, or you can use this as a way to make mealtime more interesting for dogs with problems eating. At its most advanced, an LED lights up above one of the three levers and the dog must touch the correct lever in order to get a treat. It is designed to get more and more complex during a single playing event, and after every reset, the dog has to start over. A small counter would track his overall high-score of levers in-a-row, if he could ever actually hit two in-a-row...
There are some pretty widely-accepted theories on the intelligence of dogs generally used to classify dogs. The theory identifies three types of intelligence: instinctive intelligence which describes the activities the dog was bred for, adaptive intelligence, a metric for problem-solving, and working intelligence, or the ability to learn from human interaction. I wanted to come up with a robot that could work with all three types of intelligence for some mental stimulation. Plus, tossing treats at Henry was getting a bit boring and we needed to spice it up.
The larger scheme for the idea of robotic dog toys is to have a centralized treat-delivery system that wirelessly communicates with several toys of varying design, so this 'Touch' module with the levers could just be a single toy in a group. Before I get into the specifics, check out this video of my dog Henry, playing with the finished prototype, so you can get an idea of how it works.
This is the current stage of Henry's interaction with the toy, and now I am working on him using the machine without me around.
The last and most important advice is to SUPERVISE YOUR DOG and do not leave them unattended with any of the items in this instructable. It is not meant to be an autonomous robot that wakes up and plays with your dog, but a game with which you both interact. The fun part for me was teaching Henry how to play with it, watching him systematically improve at the game, only to completely forget everything the next day.
The whole thing took about two months to complete and in the meantime I was able to spend some quality time with my dog and learn that he was pretty quick to pick up how to work the toy!
Remove these ads by
Signing UpStep 1Training the Dog
Training:
The dog has to be very comfortable around the toy for it to have any success. At the stage before I added any electronics I handfed Henry his dinner over and under and the around the beams for a few nights.
Yet he is completely motivated by food and after a few days he would be excited as soon as he saw the toy.
So first I just worked with him on touching the levers:
Since Henry already knew how to shake, I put the machine between us and tried to make him 'shake' my hand under the area of the scaffolding where the levers would go. Eventually I put my hand further and further back until he was slapping the postcards just by accident. Anytime his paw touched a postcard I gave him a tasty treat. I worked like that for a few nights and eventually just could tap the levers for him to 'touch' them. After a few times like that, all I had to do was hold the treat up and he would slap the levers.
Check it out:
You can watch Henry in the video to get a good idea of some of the challenges for this project. At one point he just double slaps two levers and leaves his paws resting on them while the Flex Sensors underneath are fully strained. So that's when I decided to separate the treat dispenser from the toy itself, so that after a tap on the levers the dog has to move a bit to collect the food.
Eventually Henry graduated to the first level of difficulty on the game. That is where we are at now because he has some trouble grasping that he should push a certain lever under whichever light was on, so I am trying to think of a way to get him to learn how to do that.
If you want to give a shot at making one yourself, there are detailed instructions and code. If you want to borrow some aspects of the module and creating something new, all I ask is that you keep it open-source!
The dog has to be very comfortable around the toy for it to have any success. At the stage before I added any electronics I handfed Henry his dinner over and under and the around the beams for a few nights.
Yet he is completely motivated by food and after a few days he would be excited as soon as he saw the toy.
So first I just worked with him on touching the levers:
Since Henry already knew how to shake, I put the machine between us and tried to make him 'shake' my hand under the area of the scaffolding where the levers would go. Eventually I put my hand further and further back until he was slapping the postcards just by accident. Anytime his paw touched a postcard I gave him a tasty treat. I worked like that for a few nights and eventually just could tap the levers for him to 'touch' them. After a few times like that, all I had to do was hold the treat up and he would slap the levers.
Check it out:
You can watch Henry in the video to get a good idea of some of the challenges for this project. At one point he just double slaps two levers and leaves his paws resting on them while the Flex Sensors underneath are fully strained. So that's when I decided to separate the treat dispenser from the toy itself, so that after a tap on the levers the dog has to move a bit to collect the food.
Eventually Henry graduated to the first level of difficulty on the game. That is where we are at now because he has some trouble grasping that he should push a certain lever under whichever light was on, so I am trying to think of a way to get him to learn how to do that.
If you want to give a shot at making one yourself, there are detailed instructions and code. If you want to borrow some aspects of the module and creating something new, all I ask is that you keep it open-source!
| « Previous Step | Download PDFView All Steps | Next Step » |
Maybe the green light goes and ultra-violet or yellow would be responsive.
Maybe an even higher level of difficulty for this module or for another type of game on a different module (for a really advanced dog) would be to have a multi-color LED and test is the dog could be trained to do different tasks to for different colors.
Thanks for the comment!