Automated tennis ball launcher for my dog

[Doc]

Hi all,

I thought I would post my thoughts/progress on designing and building an automatic tennis ball launcher for my dog. The little fella gets a bit bored during the day when we are at work, so I want to build him this so he has something to do other than push a basketball around the backyard. Any and all advice/criticism is appreciated.

Okay, so what exactly do I want this thing to do; the requirements are listed below:
1. launch a tennis ball approximately three metres and at a launch angle of 30 degrees.
2. automated so that the dog can drop the ball in and reload it himself.
3. fully enclosed so that he can't stick his paws and face in anywhere and hurt himself.
4. enough power to last four hours.
5. rechargeable power supply.

A few extra details regarding these requirements.
1. The launcher will be located on a deck which is itself situated about 1.2m above the backyard, so the ball will go a bit further than three metres. I will need to construct a stand which tilts the launcher by 30 degrees. I can make this out of wood, no problem.
2. I think it would be easy to construct this thing out of a plastic bucket with a lid. A hole can be cut in the lid where I will need to train the dog to drop the ball into (ha! Training him to do that will be the hard part!).
3. A sensor somewhere will detect whether the ball is present in the launcher. If the ball is not present, a small hatch will open exposing the hole. If the ball is present, the hatch will close so that he can't stick his face into the hole. The exit barrel can't really be enclosed but it can be made longer than his legs. The bore will need to be smaller than his head, but large enough to pass an International Tennis Federation official ball of 67mm +-1.6mm in diameter.
4. This is a bit arbitrary and can be relaxed. I suspect it may need to be due to energy demands. I also suspect that if this thing could run indefinitely the dog would eventually pass out from exhaustion (he loves chasing the ball!).
5. I am thinking solar recharging because it will be situated outside on the back deck which gets a lot of sun, though I am happy to hear other suggestions.

Conceptual design is shown in three attached images, 'ball launcher image with annotations.png', 'isometric view lid removed.png' and 'plan view lid removed.png'.

Bucket outside diameter is approximately 280mm. The basic idea is that sensors detect when the ball has been loaded. A positive ball detection triggers the motor (not shown, it will be situated underneath the unit) to rotate the hub and vanes within the bucket. The vanes rotating will launch the ball out of the barrel. Solar panels (not shown) likely mounted somewhere on the lid will charge a battery bank (not shown) situated underneath the unit next to the motor.

I scaled this unit down and 3d printed a prototype attaching a small DC motor; I was concerned that the ball wouldn't launch properly. But the prototype functioned well so that is encouraging.

I have broken this project down into three main sections: mechanics, electronics and power as shown in 'system overview.png'.

I will use an Arduino Uno to control the launcher, but if anybody has alternative suggestions I am happy to hear them.

Next steps:
- determine exit velocity required to launch ball three metres at an angle of thirty degrees.
- specify a motor (select one with excess RPM) that can satisfy the ball launch velocity, with some factor of safety to account for friction, losses etc.
- determine the energy required to perform one ball launch and correspondingly the energy required for four hours of ball launches.
- size the solar panel/battery bank to accommodate this energy requirement (this should probably be sized for cloudy days).
- design control system.

I think I'll leave it there for now. Happy to hear advice/comments.

Doc

 

[CWatters]

Looks good. Perhaps add some safety features so that the power is shut off if the dog drops a stone or stick in the bucket and it jams the rotor.

 

[phinds]

2. I think it would be easy to construct this thing out of a plastic bucket with a lid. A hole can be cut in the lid where I will need to train the dog to drop the ball into (ha! Training him to do that will be the hard part!.

Be a lot easier on the dog if you made this a funnel instead of just a hole



Just found out this is a commercial product you can buy:


it's a bit pricey though: http://www.godoggoinc.com/

 

[berkeman]

Fun project. How big/heavy is the dog? How smart and trainable is it?

If the dog is medium size and pretty smart and trainable, just make a simple spring-loaded arm with the ball holder at the top, and a release pedal at least one dog-length to the side of the launcher. Then have fun training the dog to step on the arm to push it down into horizontal launch position on the ground, drop the ball into the basket, and then step on the launch pad/button to make the ball fly out.

Then make a short video and post it on Facebook, and watch it go viral

 

[jack action]

I'm with @berkeman on this (with a small simplification): The dog drops the ball in a funnel, then the dog steps on a pedal on the side to push a spring loaded arm and once the pedal is fully depressed, the arm is released automatically, launching the ball.

 

[Doc]

Hi all,

Thanks for the comments, much appreciated!

Looks good. Perhaps add some safety features so that the power is shut off if the dog drops a stone or stick in the bucket and it jams the rotor.
Reference https://www.physicsforums.com/threads/automated-tennis-ball-launcher-for-my-dog.923244/

Yes I very much like this idea. I thought of perhaps adding in an ammeter circuit to monitor the current from the launch motor. If the motor jams-spiking the current-a switch disconnects the motor. I might add in some other error-checking/jam-clearing functionalities, but I'll think about these a bit later on.

Be a lot easier on the dog if you made this a funnel instead of just a hole
Reference https://www.physicsforums.com/threads/automated-tennis-ball-launcher-for-my-dog.923244/

Oh my god, yes! Using a funnel will make everything easier, excellent suggestion. I still want to build this thing rather than buy it though.

If the dog is medium size and pretty smart and trainable, just make a simple spring-loaded arm with the ball holder at the top, and a release pedal at least one dog-length to the side of the launcher. Then have fun training the dog to step on the arm to push it down into horizontal launch position on the ground, drop the ball into the basket, and then step on the launch pad/button to make the ball fly out.
Reference https://www.physicsforums.com/threads/automated-tennis-ball-launcher-for-my-dog.923244/

Good idea, but he's a little dog, and a little thick at that (physically and mentally). I would definitely go this route if he was bigger, thanks for the suggestion.

Work has been a bit busy lately so hopefully I can add an update on the weekend! :D

Thanks,
Doc

 

[JBA]

Here is a website calculator for your first step.

http://www.convertalot.com/ballistic_trajectory_calculator.html

 

[Doc]

Hi all,

Okay, I've worked out that the resultant launch velocity of the tennis ball needs to be about 7 m/s. In order to launch the ball, ideally, a distance of 3m, a launch velocity of approx. 3.8 m/s is required. However I am going to assume this system will have a lot of loss, so I will aim for a launch velocity of 7 m/s. All working attached 'required launch velocity.jpg'.

Now I need to convert a linear launch velocity of 7 m/s to an angular velocity and then to a spec. that I can use to select an appropriate motor (rpm). Angular velocity is linear velocity divided by radius. My radius however is not 280/2 mm (I stated in my first post that 280mm was the bucket outside diameter), see ' launch diameter.jpg'. I calculated the launch diameter to be approx. 246.5mm, the radius is of course 123.25mm.

I think that what I need to do from here is estimate the mass of the rotating central hub and the vanes: perhaps use plywood for the vanes, I might change the geometry of the hub to a triangle and use pine so it's easy to put together. Once I have the mass of the central hub plus the mass of the tennis ball I can determine the torque required to rotate everything. I can then create a few plots using stall-torque and no-load speed data from motor datasheets. I want to select a motor which has a maximum power point close to what I think my actual torque and RPM will be, see 'torque speed.jpg'.

Any advice/criticism on that last paragraph in particular would be appreciated.

Thanks,
Doc

 

[scottdave]

I saw something similar to this on TV. I think it was Outrageous Acts of Science.

 

[Tom.G]

I highly recommend a stored energy approach. Use a motor to compress a spring that drives the launcher arm/vane/whatever. That way neither the motor nor the battery need to be sized for the peak launch energy, just the average energy usage. If you need rapid turn-around, the spring can be compressed (the arm cocked) during the chase phase of the exercise. Then the device is ready as soon as the ball is returned.

COMMENT: Watching some of the videos, the anticipation of launch seems to be an important part of the game.

 

[CWatters]

Don't forget the speech synth to praise the dog for returning the ball. Isn't that why they do it?

 

[Doc]

Hi all,

No update for now, just thought I'd answer some posts.

I highly recommend a stored energy approach. Use a motor to compress a spring that drives the launcher arm/vane/whatever. That way neither the motor nor the battery need to be sized for the peak launch energy, just the average energy usage.

I suppose that I should have said originally that another half of the entire point of doing this was as a learning exercise. I haven't needed to size a motor or battery for a practical application before and thought that this would be a good opportunity. I am assume that you think that the total energy demand may be prohibitively high for a compact unit? Or is there something else that I am missing here?

I think that the next reasonable thing to look at would be the energy requirement to run the thing for four hours. If it can be done with a battery that is relatively compact, then okay. But if not I may have to rethink the entire design and go for a spring launch system.

Don't forget the speech synth to praise the dog for returning the ball. Isn't that why they do it?

Ha! Either that or I design a smaller secondary dog-treat container and launch cycle to be one launch out-of-phase with the tennis ball!

Thanks,
Doc

 

[jack action]

I am assume that you think that the total energy demand may be prohibitively high for a compact unit? Or is there something else that I am missing here?

It's not the amount of energy that is worrisome, it's how fast it needs to be delivered: That's power (energy over time).

For example, let's say your ball has a mass of 60 g and the initial velocity is 7 m/s. The net energy required by your machine is $\frac{1}{2}mv^2$ or 1.47 J (= 1/2 * 0.060 * 7²).

If you have a spring loaded machine that loads the spring while the dog is chasing the previously thrown ball and that the dog takes 5 s before coming back with the ball, you can use a motor producing as low as 0.294 W (= 1.47 / 5). If you go with your machine that starts when the ball drops in (which should turn at around 500 rpm to produce the required 7 m/s), you will have 1/500 min to do that 1 rev before the ball exits, or 0.12 s. So you need a motor that produces at least 12.25 W of power (= 1.47 / 0.12). And that is only considering the mass of the ball, not including the inertia of the rotating machine itself or the friction losses.

And if you want to start your machine from rest every time, you have to think that you will need to go from 0 to 500 rpm within 1 rev. You need to be able to reach such level of acceleration.

 

[jim hardy]

 

[Doc]

It's not the amount of energy that is worrisome, it's how fast it needs to be delivered: That's power (energy over time).

For example, let's say your ball has a mass of 60 g and the initial velocity is 7 m/s. The net energy required by your machine is $\frac{1}{2}mv^2$ or 1.47 J (= 1/2 * 0.060 * 7²).

If you have a spring loaded machine that loads the spring while the dog is chasing the previously thrown ball and that the dog takes 5 s before coming back with the ball, you can use a motor producing as low as 0.294 W (= 1.47 / 5). If you go with your machine that starts when the ball drops in (which should turn at around 500 rpm to produce the required 7 m/s), you will have 1/500 min to do that 1 rev before the ball exits, or 0.12 s. So you need a motor that produces at least 12.25 W of power (= 1.47 / 0.12). And that is only considering the mass of the ball, not including the inertia of the rotating machine itself or the friction losses.

And if you want to start your machine from rest every time, you have to think that you will need to go from 0 to 500 rpm within 1 rev. You need to be able to reach such level of acceleration.

This makes perfect sense thank you, and I see now that I did not approach this problem properly. I have made design decisions without adequately understanding what I want to do.

Perhaps the way I should have thought about this is:
- I want to do x, x requires energy.
- How much energy is required to do x once?
- What are DIFFERENT ways to do x?
- How quickly will these different ways operate? This will give an approximate time associated with each option.
- Divide the energy required to do x by the time associated with each option to get Power required.
- From here it should be clear which options will require higher power (and correspondingly either high voltage and/or current) or lower power.

I think I'll go back to the start and do this step first.