I've been working on reinventing a trainer system. I have a prototype and it is working rather nicely for me; given the recent thread about Kinect I thought I'd share:
The system consists of the following components:
1. Laser module attached to the pistol or rifle, beam directed towards the target but not at it, such that the beam is occluded by the rifle/pistol and does not interfere with the sight picture.
2. A camera captures the laser beam spot (may be placed close to the target when using short focal length lens, or next to the shooter when using powerful lens)
3. A microphone captures the shot event (regular microphone works fine for quiet environment; a contact piezoelectric microphone can be attached to the pistol/rifle for use in a loud environment).
4. The audio and video feeds are processed by the software to calculate / display / analyze the shot trace.
5. On the back-burner: A second camera is oriented towards the shooter and captures the position, orientation, and motion of the rifle/pistol, allowing reconstruction of the complete sight picture (independent movement of the front and rear sights relative to the target, as viewed by the shooter, in addition to the trace on the target as captured by the first camera).
Screengrab of the system in action, showing review of a few shots: http://www.youtube.com/watch?v=m91tzH3GOpI (the original is at 60fps and looks a lot smoother than compressed youtube video)
The cost of hardware (camera, lens, tripod, microphone, cabling, laser) is around $450-$500.
Do you think there would be any demand for this sort of thing?
Home Trainers: A challenger appears.
Moderators: pilkguns, m1963, David Levene, Spencer, Richard H
Home Trainers: A challenger appears.
- Attachments
-
- Beam capture
-
- Camera
-
- Pistol with Laser
-
- Super Jerk
-
- OK shot
-
silentfury214
- Posts: 186
- Joined: Thu Jun 24, 2010 9:07 am
- Location: Tennessee
Richard and silentfury214,
At present these are the hurdles that affect Mac support:
1) The hardware I used is not supported on Mac; I'm going to search around to see if I can find an affordable USB camera with Mac support, at least 640x480@60fps/uncompressed, preferably accepting C/CS-mount lens. If you are aware of such a beast, please let me know.
2) Libraries I rely upon require training in dark arts to run properly on Mac.
3) I don't have a Mac and no experience developing for it (this one is solveable).
Brian,
The software at present state is a proof-of-concept. The development essentially stopped once I got basic functionality to work, as now I suddenly have much less free time : ). At present, the functionality is very basic: after the beam leaves the aiming area, the software goes into review mode and loops the last shot. In terms of training, I find "historical" traces less helpful, as the mental association between what I was doing at the time and how it is reflected by the trace is lost by that time. Alternatively, if one wants to have a birds-eye view of a training session I think that a grid of trace snapshots (e.g. 6x10 grid) would be useful, rather than having a video of each - what do you think? I also may support dumping the raw coordinates of the trace to file and have the user run wild with it.
paulo,
I'm not sure what's the best way to proceed at this point: whether to start a company and make it a supported product, or make it a "donateware" project with limited software-only support, and in that case whether or not have it open-sourced. Hence I'm doing a little "market research" : )
Any thoughts are welcome.
At present these are the hurdles that affect Mac support:
1) The hardware I used is not supported on Mac; I'm going to search around to see if I can find an affordable USB camera with Mac support, at least 640x480@60fps/uncompressed, preferably accepting C/CS-mount lens. If you are aware of such a beast, please let me know.
2) Libraries I rely upon require training in dark arts to run properly on Mac.
3) I don't have a Mac and no experience developing for it (this one is solveable).
Brian,
The software at present state is a proof-of-concept. The development essentially stopped once I got basic functionality to work, as now I suddenly have much less free time : ). At present, the functionality is very basic: after the beam leaves the aiming area, the software goes into review mode and loops the last shot. In terms of training, I find "historical" traces less helpful, as the mental association between what I was doing at the time and how it is reflected by the trace is lost by that time. Alternatively, if one wants to have a birds-eye view of a training session I think that a grid of trace snapshots (e.g. 6x10 grid) would be useful, rather than having a video of each - what do you think? I also may support dumping the raw coordinates of the trace to file and have the user run wild with it.
paulo,
I'm not sure what's the best way to proceed at this point: whether to start a company and make it a supported product, or make it a "donateware" project with limited software-only support, and in that case whether or not have it open-sourced. Hence I'm doing a little "market research" : )
Any thoughts are welcome.
Understood (and congrats on having less free time). :)AAlex wrote:
Brian,
The software at present state is a proof-of-concept. The development essentially stopped once I got basic functionality to work, as now I suddenly have much less free time : ). At present, the functionality is very basic: after the beam leaves the aiming area, the software goes into review mode and loops the last shot. In terms of training, I find "historical" traces less helpful, as the mental association between what I was doing at the time and how it is reflected by the trace is lost by that time. Alternatively, if one wants to have a birds-eye view of a training session I think that a grid of trace snapshots (e.g. 6x10 grid) would be useful, rather than having a video of each - what do you think? I also may support dumping the raw coordinates of the trace to file and have the user run wild with it.
It's been 3 years since I spent time with RIKA, but knowing how I train and shoot means that *I* need to be able to review prior shots. When things are going well/right for me I tend to not pause to view data collected instead focusing on reinforcing the positive shot process as many times as possible. But I do want to go back and take a look at the data to see if I can find common themes (I learned that 90%+ of my good shots happen within 2 seconds of getting on the target with the RIKA).
Finding out how others use current systems (and their perceived limitations of those systems) would be how I'd proceed IF I were to proceed.
As for pricing, one other product I use (software for working on Volkswagen/Audi vehicles) sells just the hardware with no software support AND they sell software support (obviously making a "package deal" appealing). That means they cover the used market too (if someone wants support).
Very cool!!! The big advantage your system could have is that the bulk of the tech needed for the system to work many of us already have at home or can purchase very cheap. The way I see this possibly progressing is you sell the software and maybe the laser, and your customers provide the camera and mic. I'm not sure of what compatibility issues the various camera types might cause.
Also, your estimate of $450-500 might even be a touch high if you were to experiment with using a web cam close up to the laser target surface. If all worked well, you can get a webcam for $15, plus 50ft of cable for another $40ish to go with a camera stand and mic the customer already has, and ~$50ish (?) laser gets you to $100. At that point, you need a mic ($10 if not hidden somewhere around the house), laser with mount (likely sold by you), and your software.
In terms of demand, if you can do the following, you'll have a big winner:
- make it compatible with a range of web cams
- make it plug and play such that you can head to the range and plug it all in to a laptop
- make the software operate like noptel/rika (shot string recording, group size calculations, scoring to the tenth of a point)
- make it able to operate while dry firing
Also, your estimate of $450-500 might even be a touch high if you were to experiment with using a web cam close up to the laser target surface. If all worked well, you can get a webcam for $15, plus 50ft of cable for another $40ish to go with a camera stand and mic the customer already has, and ~$50ish (?) laser gets you to $100. At that point, you need a mic ($10 if not hidden somewhere around the house), laser with mount (likely sold by you), and your software.
In terms of demand, if you can do the following, you'll have a big winner:
- make it compatible with a range of web cams
- make it plug and play such that you can head to the range and plug it all in to a laptop
- make the software operate like noptel/rika (shot string recording, group size calculations, scoring to the tenth of a point)
- make it able to operate while dry firing
Make replay playable in slow motion, and with a pause option.
The use of different colors during a shot used by RIKA is very useful for visualization of the different phases of a shot.
Folders for saving multiple users, as mentioned, will make it appealing to clubs.
How would you attach the laser to an IZH, one of the most used guns in training and competition from my point of view, and keep the all system under the price of that same gun 350-390?
As mentioned an affordable usb camera could be a cheaper approach, don't know if that will give you a good trace for computer analysis.
Talk to people like PilkGuns and assess the size of the market, they could better give you an idea of the US market size for a product like this, or even point you towards venture capital.
These people http://www.centershot.com/ seem to be another interesting contact for you, they deal with the Scatt, and could be a way to reach the source and sale them the prototype.
A college with a shooting team could be another avenue for you to have the software developed.
Here is a place that wont bight, but I would consider contacting, they do alot of science fiction stuff for the army, this could be something they could be interested in, http://www.sarnoff.com/ , don't be intimidated all great ideas start small, but they are a giant guided towards real life application of visual science break throughs, these are the guys that make the cameras that see behind buildings, incredible stuff.
The all shooting community has been waiting for an affordable product like this for a long time.
Good luck lets us know what you decide.
The use of different colors during a shot used by RIKA is very useful for visualization of the different phases of a shot.
Folders for saving multiple users, as mentioned, will make it appealing to clubs.
How would you attach the laser to an IZH, one of the most used guns in training and competition from my point of view, and keep the all system under the price of that same gun 350-390?
As mentioned an affordable usb camera could be a cheaper approach, don't know if that will give you a good trace for computer analysis.
Talk to people like PilkGuns and assess the size of the market, they could better give you an idea of the US market size for a product like this, or even point you towards venture capital.
These people http://www.centershot.com/ seem to be another interesting contact for you, they deal with the Scatt, and could be a way to reach the source and sale them the prototype.
A college with a shooting team could be another avenue for you to have the software developed.
Here is a place that wont bight, but I would consider contacting, they do alot of science fiction stuff for the army, this could be something they could be interested in, http://www.sarnoff.com/ , don't be intimidated all great ideas start small, but they are a giant guided towards real life application of visual science break throughs, these are the guys that make the cameras that see behind buildings, incredible stuff.
The all shooting community has been waiting for an affordable product like this for a long time.
Good luck lets us know what you decide.
Thanks for the feedback.
Dry-firing is already supported - the microphone picks up the click just fine, even with quieter Morini electronic triggers. Is LP10e click any quieter?
Calculating and recording shot values, etc. should be rather straightforward.
Hardware-wise, I tested the top of the line consumer-grade webcams from Logitech, Microsoft, and Philips and found all of them lacking, while the entry-level industrial-grade machine-vision cameras begin at close to $1k, and there's almost no products that fill the gap. I'm currently using $400 camera from ImagingSource, but I understand that this may be a hard sell.
I'm going to try Sony PS3Eye camera next, which, if successful, would bring the total hardware cost down considerably.
Dry-firing is already supported - the microphone picks up the click just fine, even with quieter Morini electronic triggers. Is LP10e click any quieter?
Calculating and recording shot values, etc. should be rather straightforward.
Hardware-wise, I tested the top of the line consumer-grade webcams from Logitech, Microsoft, and Philips and found all of them lacking, while the entry-level industrial-grade machine-vision cameras begin at close to $1k, and there's almost no products that fill the gap. I'm currently using $400 camera from ImagingSource, but I understand that this may be a hard sell.
I'm going to try Sony PS3Eye camera next, which, if successful, would bring the total hardware cost down considerably.
Is this kind of stuff out the question, because of the same reasons as computer cameras?
http://cgi.ebay.com/Super-Mini-1-3-Colo ... 335d1a6a65
http://cgi.ebay.com/6-0-mm-1-3-CS-MOUNT ... 5add686d8f for the fixed iris
http://cgi.ebay.com/ULTRAK-KC552BCN-CCT ... 35aef6c2d7
http://cgi.ebay.com/COSTAR-CCC3400-CCTV ... 3a61d7298c
http://cgi.ebay.com/sony-ccd-DVR-DVR-ca ... 29fed5ff5e
http://cgi.ebay.com/Super-Mini-1-3-Colo ... 335d1a6a65
http://cgi.ebay.com/6-0-mm-1-3-CS-MOUNT ... 5add686d8f for the fixed iris
http://cgi.ebay.com/ULTRAK-KC552BCN-CCT ... 35aef6c2d7
http://cgi.ebay.com/COSTAR-CCC3400-CCTV ... 3a61d7298c
http://cgi.ebay.com/sony-ccd-DVR-DVR-ca ... 29fed5ff5e
paulo,
Thanks for additional research.
Analog cameras will require additional hardware (frame grabber), and their specs and performance is targeted for surveillance applications rather than machine vision.
Having experimented with multiple cameras, I found out the following:
1) Resolution is immaterial, as long as higher than 320x240.
2) Frame-rate at least 60 fps (PS3Eye works at 60 fps interlaced, but I think it should work).
3) Manual or API control over focus, iris, gain, and exposure.
4) USB2.0 or IEEE1394 interface.
Unfortunately, there are only few products that satisfy the above and don't have obscene price-tags.
An HDTV camcorder operating at 60fps would work great, but they don't have PC streaming capabilities; it is also unlikely that it would be possible to control the camera in real-time via PC.
Thanks for additional research.
Analog cameras will require additional hardware (frame grabber), and their specs and performance is targeted for surveillance applications rather than machine vision.
Having experimented with multiple cameras, I found out the following:
1) Resolution is immaterial, as long as higher than 320x240.
2) Frame-rate at least 60 fps (PS3Eye works at 60 fps interlaced, but I think it should work).
3) Manual or API control over focus, iris, gain, and exposure.
4) USB2.0 or IEEE1394 interface.
Unfortunately, there are only few products that satisfy the above and don't have obscene price-tags.
An HDTV camcorder operating at 60fps would work great, but they don't have PC streaming capabilities; it is also unlikely that it would be possible to control the camera in real-time via PC.
-
David Levene
- Posts: 5617
- Joined: Mon Mar 01, 2004 12:49 pm
- Location: Ruislip, UK
-
Xpermen (Spain)
Re: Home Trainers: A challenger appears.
Congratulation for your big effort in making the hardware more affordable, nevertheless you should not forget that a powerful software will make the difference. I do not know the RIKA system but I have used Noptel and Scatt. SCATT software, in my opinion, is the best, and could be a good reference: shooting parameters are adjustable, data can be analyzed from several points of view ... Nevertheless, the interface and ease of use can be improved, so, this is the big challenge
I dont want to clutter up this thread with 'great idea messages' but I suspect there are a lot of people like me out here who would go for a system like this - I wont stretch to the current systems but would almost certainly go for something at the level of costs you are floating.
Particularly (because i cant stop fiddling/experimenting) if there was a degree of flexibility in the composition/assembly/upgrading options.
Good luck with the project
John
PS and make it easily available in the UK please!!
Particularly (because i cant stop fiddling/experimenting) if there was a degree of flexibility in the composition/assembly/upgrading options.
Good luck with the project
John
PS and make it easily available in the UK please!!
David-
Yes, there was a touch of humor intended there. Perhaps more than a touch. But– a few years ago, I would have been the only person to suggest Mac compatibility. Now, two others beat me to it. That tells you that even within our small community of shooters, there is a GROWing number of people, and hence a growing percentage, who would buy this thing, probably immediately, if it were Mac compatible.
Apple's market share is closer to 7% depending on whose stats you use, but the point is, the user base is growing faster than the industry itself. So there is an opportunity here for a tech person to grab a growing niche market.
-Larry
Yes, there was a touch of humor intended there. Perhaps more than a touch. But– a few years ago, I would have been the only person to suggest Mac compatibility. Now, two others beat me to it. That tells you that even within our small community of shooters, there is a GROWing number of people, and hence a growing percentage, who would buy this thing, probably immediately, if it were Mac compatible.
Apple's market share is closer to 7% depending on whose stats you use, but the point is, the user base is growing faster than the industry itself. So there is an opportunity here for a tech person to grab a growing niche market.
-Larry
Except for the inconvenience of 24 VAC power, here are a few options for reasonable priced cameras, BNC video plug easy to adapt to RCA, also I don't think they have a mic
http://www.nellyssurplus.com/servlet/th ... Categories
and power supply
http://www.nellyssurplus.com/servlet/th ... wer/Detail
Now for the capture there are this gadgets that capture video trough a usb, let me see if I find that stuff, voila
http://www.amazon.com/s/?ie=UTF8&keywor ... ay463ry8_e
I think we got that solved, what's next?!
http://www.nellyssurplus.com/servlet/th ... Categories
and power supply
http://www.nellyssurplus.com/servlet/th ... wer/Detail
Now for the capture there are this gadgets that capture video trough a usb, let me see if I find that stuff, voila
http://www.amazon.com/s/?ie=UTF8&keywor ... ay463ry8_e
I think we got that solved, what's next?!
One feature I wish my SCATT had--enough that I could be swayed to buy something in the $500 range-- is better data export. I'd like to be able to get a dump of either the raw points of the trace (I can write a script to turn than into the parameters I want to analyze) or all the parameters the SCATT gives like: distance from center of the aiming dot, length, length during shot phases, % in 10 and 10.5 rings, etc.
What I'm really interested in is being able to answer this question: do my shots follow a generally Gaussian distribution (extreme events occur with the frequency predicted by the normal distribution) --in which case I need to make the variance smaller to keep them in the 10--or are "bad" shots over-represented, like they come from a different distribution--in which case I should focus on figuring out how to reject those shots before they become 9s.
Another feature that would probably get me to part with my money: if I could set the system up outside at my local range and then shoot for real. (With a smallbore or even a centerfire gun) I would be able to set the screen up closer than my actual target with the layout of my range. I would really like to be able to use it as an over-fancy "ball and dummy drill"
Best of luck in getting this thing going!
What I'm really interested in is being able to answer this question: do my shots follow a generally Gaussian distribution (extreme events occur with the frequency predicted by the normal distribution) --in which case I need to make the variance smaller to keep them in the 10--or are "bad" shots over-represented, like they come from a different distribution--in which case I should focus on figuring out how to reject those shots before they become 9s.
Another feature that would probably get me to part with my money: if I could set the system up outside at my local range and then shoot for real. (With a smallbore or even a centerfire gun) I would be able to set the screen up closer than my actual target with the layout of my range. I would really like to be able to use it as an over-fancy "ball and dummy drill"
Best of luck in getting this thing going!
Oh, and one other thought...
Have you considered using an IR laser-camera combination (might be able to use the current camera with a couple of filters to isolate the IR band)? That would allow you to put the laser target closer to the "real" target, which would minimize the effect of parallax errors caused by moving the position slightly compared to the target. (if the laser isn't in the same position in space from one shot to the next, even though it's pointed at the target, the system would be more likely to see it as aimed somewhere else.) Yes, that would help you keep your position solid, but it would also cause phantom drifts in the displayed hit location which could cause a lot of frustration trying to eliminate it.
Have you considered using an IR laser-camera combination (might be able to use the current camera with a couple of filters to isolate the IR band)? That would allow you to put the laser target closer to the "real" target, which would minimize the effect of parallax errors caused by moving the position slightly compared to the target. (if the laser isn't in the same position in space from one shot to the next, even though it's pointed at the target, the system would be more likely to see it as aimed somewhere else.) Yes, that would help you keep your position solid, but it would also cause phantom drifts in the displayed hit location which could cause a lot of frustration trying to eliminate it.