Combat Robotics

[W. Rabbit]

Couldn't find a better place for this thread, but before anyone decides to turn on their moron mode, this is a serious thread meant for serious discussion as part of a real engineering project I am beginning work on in the area of combat robotics.

For reasons of protecting my work I do not want to share too many details about the conception/goal of this project to the general community. I will consider an NDA-based approach in the future if there is enough serious interest from anyone here.

I do have a rough design goal in mind and am currently doing a lot of patent searches and IEEE reviews to find more information. This BS thread is part of that search.

I would like this thread to serve a few functions:

• Provide a way for me to distribute RFIs to a relatively intelligent, educated community of hand to hand fighters, both amateur and professional.
• To receive feedback on ideas, applications, or design issues I may come across.
• To better apply my understanding of electrical/computer/robotic engineering with the physical realities/limitations of combat.
• To potentially gain access to folks who could help me review past engineering work done out of Japan, since I do not speak or read Japanese.

Note: none of my research is classified, publicly funded, or involves the military. I am specifically interested in robotic applications for human hand to hand combat.

Thanks.

[Rene "Zendokan" Gysenbergs]

I'm seeing a bottleneck with the software even if you could build a lightweight drone and even then it won't be 'mobile' to travel from one destination to another.

If the drone doesn't get visual input (the sheer mass of data processing would be staggerant), it would mean that the human opponent should wear a bodysuit with build in sensors.

If we use English Boxing as the start Combat Sport (so no kicking and grappling since those two tasks are even more difficult to achieve) the data would be tremendous:
- All the data that is needed to control the drone (Boxing techniques, balance, movement, walking)
- All the data coming from the bodysuit to pinpoint the user and his actions (can be easely upto 50 sensors)
- Constant recalculation because of the constant changing of the situation between human opponent and drone

This would mean that the drone itself would sent and receive data from a computersystem (maybe even a midframe) that would be standing besides the ring or tatami. The system would be to heavy/big to build into the drone.

There's no program that can proces so much data at a max cycle rate of 0.2 seconds constantly for a period of 5 minutes.

Let alone the data that you would get when you add kicking to it or grappling (minimum 50 sensors extra on the drone's body alone) so that the human doesn't get hurt.

[goodlun]

I think we need to really start with applications what exactly do you want your h2h robot accomplish?

[Rene "Zendokan" Gysenbergs]

I'm going to play along, even if I don't believe this at all.

First it has to be a "from distance controlled" drone without any form of A.I.:
- A.I. technology is still in its infantiel state
- an independant operating battledroid will pose a danger in the long run (hacks, software bugs)

Second, since it will be a controlled drone it will need:
- a nonhackable connection with the homebase that does datatransfers in real-time
- a non blockable connection at all times and places (for drone operating commands)
- datatransfers will exist out of visual and audio (speach and listening) data since it will be a one-on-one connection between a human operator and the drone and drone operating data (movement, steering)

Third, weaponsystems should not be integrated into the drone body:
- keeps the drone body relative compact and light
- makes it easier to armour the drone so that the control and communication technology will be better protected
- prevents from having an additional weakness inside the frame (anology: fireworks on a open hand = burn wounds, fireworks in a closed fist = bye bye hand)

Fourth: Most important!
- If you really believe that Militairy contracters haven't/aren't developing and researching this already than you are really delusional!

[goodlun]

The biggest problem I see with this is anamorphic thinking.
Ideally something like the Taser shockwave(http://www.youtube.com/watch?v=bZfYo...eature=related) mounted to say an EOD drone is most likely going to prove more effective in the end.

[W. Rabbit]

I think we need to really start with applications what exactly do you want your h2h robot accomplish?

Without giving up my design goal (which I think is relatively unseen in the commercial sector, but has great potential for interest), I'd say the following are basic pieces of the puzzle:

• Direct human physical interaction
• Force feedback
• Voice commanded
• Mobile
• Non-visual based (meaning no photoreceptors, "machine vision" etc)

[Kovacs]

((cross post))

[W. Rabbit]

First it has to be a "from distance controlled" drone without any form of A.I.:
- A.I. technology is still in its infantiel state
- an independant operating battledroid will pose a danger in the long run (hacks, software bugs)

I don't think those would apply, this wouldn't be a "production" grade military device, but I could see it being used in a lot of different training environments: military, commercial, and personal. It would really not need any advanced AI, and would be feedback based or run a specific program.

Second, since it will be a controlled drone it will need:
- a nonhackable connection ...

I like the way you think, but I don't think remote security will be a concern, at least not until Skynet becomes self aware and somehow decides to conscript my robots. :)

Third, weaponsystems should not be integrated into the drone body:
- keeps the drone body relative compact and light
- makes it easier to armour the drone so that the control and communication technology will be better protected
- prevents from having an additional weakness inside the frame (anology: fireworks on a open hand = burn wounds, fireworks in a closed fist = bye bye hand)

While this wouldn't be a weaponized platform, your ideas here made me think that this thing will definitely require special protective components. I can see it taking a lot of wear and tear.

Fourth: Most important!
- If you really believe that Militairy contracters haven't/aren't developing and researching this already than you are really delusional!

They most definitely are but those ones will carry weapons into the field, are mostly classified, and unaffordable to most people. Bomb defusing robots are definitely one source of my inspiration here but I am actually specifically interested in robotics with applications for fighting, fight training, conditioning, and sparring practice, with a price tag within the $5,000-$10,000 retail price range opening it to most consumers/gyms, etc.

Thanks for the input.

[goodlun]

I guess the real question then is more along the lines is does it have to be able to "grapple", does it need the ability to man handle a person and move them from point A to Point B with out their compliance?

[Kovacs]

Without giving up my design goal (which I think is relatively unseen in the commercial sector, but has great potential for interest), I'd say the following are basic pieces of the puzzle:

• Direct human physical interaction
• Force feedback
• Voice commanded
• Mobile
• Non-visual based (meaning no photoreceptors, "machine vision" etc)

I've got a sneaky feeling you want to build a sparring droid of some sort. Sounds tough but I imagine it would be a good seller of it could be made cheaply enough.

Edit: gah, cross post again.