Sensor Tip
Giving your robot whatever sensors ( the more the merrier ) helps bring your creation to life. However there is a problem regards touch, most robots are small and light (lack inertia and mass ) so this means you need a small feather lite touch sensor or the only thing you will be able to detect is crashing into a solid object with undesired force (and insufficient reaction time)
Solution
Take a piece of solder ( 2 to 4 cm long ) and fold it into a W shape, sellotape this to one surface and to the other surface ( foam of some sort) place two small squares of metal almost touching. ( punch a hole through the foam and solder wires to the back of the metal contacts)
Why it works so well !.....The solder is flexible so when its pushed into the small metal contacts it starts to bend into a 'close fit shape'. Tiny flat electrical contacts being what they are don't make a complete circuit even if they are laying onto of each other. You have hard flat metal resting on hard flat metal but still no current flow. However add a tiny amount of pressure and you've closed the circuit. Enough electricity can now flow to light an led or trigger the gate on some chip or transistor.
HAL9000 thinking machine all the bells & whistles
Re: HAL9000 thinking machine all the bells & whistles
Sensor continued.
How sensitive is the above method, the W shaped solder bridge ?
I dropped a 2 inch wood screw from 4 inches height onto the sandwich. ( foam/w+contacts/foam ) and the led i had wired up flickered to the bounceing of the screw !...the size of the sensor in question is just over 1cmx1cm but could be triggered over an area approx 4cmx4cm.
A typical switch or contact usually requires 100s of grams of force especially when its under a layer of foam or some other stuff. Its also very hard to get a typical micro-switch (with lever) to trigger unless the force is applied at just the right angle. While they might only cost a couple of quid if you want/need dozens or 100s of them the cost soon adds up. Using the method i described above you quickly and very easily make a dozen sensors for the price of a single micro-switch.
The more sensors your robot has the better and if you can duplicate or triplicate the sensors then not only can cross reference the results but you are no longer at risk from weird behavior or a single point of failure. The effective intelligence of your robot is closely linked to the number and diversity of sensors it has. Once you get over a dozen it starts to become interesting, once you get over 100 it starts to become lifelike and if you can exceed 1000 sensors its sensory instincts are not that far behind a real creature and that's when the programming side of things starts to become fun and very rewarding.
How sensitive is the above method, the W shaped solder bridge ?
I dropped a 2 inch wood screw from 4 inches height onto the sandwich. ( foam/w+contacts/foam ) and the led i had wired up flickered to the bounceing of the screw !...the size of the sensor in question is just over 1cmx1cm but could be triggered over an area approx 4cmx4cm.
A typical switch or contact usually requires 100s of grams of force especially when its under a layer of foam or some other stuff. Its also very hard to get a typical micro-switch (with lever) to trigger unless the force is applied at just the right angle. While they might only cost a couple of quid if you want/need dozens or 100s of them the cost soon adds up. Using the method i described above you quickly and very easily make a dozen sensors for the price of a single micro-switch.
The more sensors your robot has the better and if you can duplicate or triplicate the sensors then not only can cross reference the results but you are no longer at risk from weird behavior or a single point of failure. The effective intelligence of your robot is closely linked to the number and diversity of sensors it has. Once you get over a dozen it starts to become interesting, once you get over 100 it starts to become lifelike and if you can exceed 1000 sensors its sensory instincts are not that far behind a real creature and that's when the programming side of things starts to become fun and very rewarding.
Re: HAL9000 thinking machine all the bells & whistles
Bio realistic Robot
This was so good that the first time i saw it ( carpet only video) i actually thought it was fake. Very inspiring. No doubt some well written code driving it all.
http://www.youtube.com/watch?v=-vVblGlIMgw
Tech tip for robots with cameras
If your robot has a camera the it can read (see the light from ) LED attached to simple sensors. There are plenty of videos on you tube that show you how to build a DIY touch table. The same technique can be used to upgrade and enhance any robot equipped with a camera.
Additional tip. If an led is placed some way from the camera and an object is impinging on your robots body space then fact that you can no longer see that led, ie its been covered by a cloth or some artifact, can allow for cross referencing...eg its not just dark,or a grey out condition but something is in between the camera on your robot and the part of the body you placed that led. Leds are bright and are easy to pick out using simple FreeBASIC code. Cross referencing sensory inputs is a surefire way to import 'external' realities into your robots mind for increased situational awareness.
Examples i actually deploy on my robot/robots
Led attached to a tilt switch.
Led attached to a sound to light switch.
Led attached to a knock/vibration sensor.
Led attached to a light dependent resistor. ( great for determining light levels in all directions not just the direction your robots camera is facing)
A strip of LEDs arranged so that they appear along the bottom edge of your field of view is a good way to go. Also you can encapsulate those LED's in tiny light tubes. This means you can have a thin dark strip with pinpoints of light rather than a lens flare effect which may make the rest of the field of view difficult to see.
There is code elsewhere on this forum that will allow you to capture the feed from a webcam. An afternoon of experimentation and you will be good to go.
This was so good that the first time i saw it ( carpet only video) i actually thought it was fake. Very inspiring. No doubt some well written code driving it all.
http://www.youtube.com/watch?v=-vVblGlIMgw
Tech tip for robots with cameras
If your robot has a camera the it can read (see the light from ) LED attached to simple sensors. There are plenty of videos on you tube that show you how to build a DIY touch table. The same technique can be used to upgrade and enhance any robot equipped with a camera.
Additional tip. If an led is placed some way from the camera and an object is impinging on your robots body space then fact that you can no longer see that led, ie its been covered by a cloth or some artifact, can allow for cross referencing...eg its not just dark,or a grey out condition but something is in between the camera on your robot and the part of the body you placed that led. Leds are bright and are easy to pick out using simple FreeBASIC code. Cross referencing sensory inputs is a surefire way to import 'external' realities into your robots mind for increased situational awareness.
Examples i actually deploy on my robot/robots
Led attached to a tilt switch.
Led attached to a sound to light switch.
Led attached to a knock/vibration sensor.
Led attached to a light dependent resistor. ( great for determining light levels in all directions not just the direction your robots camera is facing)
A strip of LEDs arranged so that they appear along the bottom edge of your field of view is a good way to go. Also you can encapsulate those LED's in tiny light tubes. This means you can have a thin dark strip with pinpoints of light rather than a lens flare effect which may make the rest of the field of view difficult to see.
There is code elsewhere on this forum that will allow you to capture the feed from a webcam. An afternoon of experimentation and you will be good to go.
Re: HAL9000 thinking machine all the bells & whistles
Key Tech tip for WALKING ANDROIDS
I have been carrying out numerous experiments on an almost daily basis trying to figure out cheap n cheerful cheats ref walking androids. Walking anything is hard, especially so if your over 4feet tall and slim like a real human being. With this in mind the following cheat works really well.
1) Have your realistic looking android feet 5 to 10% over-sized ( you can get away with that and still keep it looking bio realistic)
2) Place large weights in the toe and heel, this helps with leg swing AND tends to make the android self right if its only 5 degrees from vertical. ( assumes your android is relatively light compared to the feet weights ) If you can shunt 25% of the total weight in the feet or lower leg, ie batteries this also helps....yes its cheating very badly but it works !.
Most falls start with an unwanted wobble of just a few degrees so concentrating /deliberately adding weight to the feet will give you a much greater margin of safety/error. Walking androids is as hard as it gets so don't be ashamed of pulling every trick that you can just to cross the floor.
*if you had a simple on off pressure sensor on each corner of the foot you could attach 4x2 leds in such a way they can be seen by the eye camera. ( disguise hint sunglasses or a few inches of fiber optic on the bridge of the nose )...little LED sparkles help prevent big falls :-p
I have been carrying out numerous experiments on an almost daily basis trying to figure out cheap n cheerful cheats ref walking androids. Walking anything is hard, especially so if your over 4feet tall and slim like a real human being. With this in mind the following cheat works really well.
1) Have your realistic looking android feet 5 to 10% over-sized ( you can get away with that and still keep it looking bio realistic)
2) Place large weights in the toe and heel, this helps with leg swing AND tends to make the android self right if its only 5 degrees from vertical. ( assumes your android is relatively light compared to the feet weights ) If you can shunt 25% of the total weight in the feet or lower leg, ie batteries this also helps....yes its cheating very badly but it works !.
Most falls start with an unwanted wobble of just a few degrees so concentrating /deliberately adding weight to the feet will give you a much greater margin of safety/error. Walking androids is as hard as it gets so don't be ashamed of pulling every trick that you can just to cross the floor.
*if you had a simple on off pressure sensor on each corner of the foot you could attach 4x2 leds in such a way they can be seen by the eye camera. ( disguise hint sunglasses or a few inches of fiber optic on the bridge of the nose )...little LED sparkles help prevent big falls :-p
Re: HAL9000 thinking machine all the bells & whistles
Been doing a ridiculous amount of R&D ref realistic android eyes. For now realistic tilt n pan mechanisms fit for androids or small robots are basically non existent.
The day you can buy a USB powered fully functional android eye for $200 (or less) a pair will be the day that coders stop sniffing the solder and start developing 'meaningful' vision software fit for every day real world applications.
Plug n play android eyes ? We are not there yet BUT im getting pretty close to a production worthy system.
Yet again the astronomers are leading the way. DIY eyeball camera hack.
http://www.youtube.com/watch?v=iarXeMkIJ2E
The day you can buy a USB powered fully functional android eye for $200 (or less) a pair will be the day that coders stop sniffing the solder and start developing 'meaningful' vision software fit for every day real world applications.
Plug n play android eyes ? We are not there yet BUT im getting pretty close to a production worthy system.
Yet again the astronomers are leading the way. DIY eyeball camera hack.
http://www.youtube.com/watch?v=iarXeMkIJ2E
Re: HAL9000 thinking machine all the bells & whistles
Finally solved the humanoid eye mech. I literately had to sit down and write a flow chart of the various issues, solutions and issues that stemmed from each solution. I ended up with a truly unique mechanism, something id never seen before. While their are several basic solutions that are easily found on the web they invariably lack one or more key features, or worse still the way they are designed precludes other capabilities found in real eyes.
Some close up shots of the HRP android dancing.
http://www.youtube.com/watch?v=bLTdM8Y9nj8
OpenHRP3 (Open Architecture Human-centered Robotics Platform version 3)
http://www.openrtp.jp/openhrp3/en/about.html
Spent most of this weekend shopping for womens clothes, wigs and makeup etc....my cover story was shooting an indie film :-p ...that worked most of the time
Miss 'C' will be a tiny bit taller than HRP-4C , kind of like Goldilocks with attitude, lol
http://en.wikipedia.org/wiki/HRP-4C
She will also be 99% identical to a real human bodyform, HRP is close, close enough for most intents and purposes but she will never be able to move with the same grace as a real human due to the shortage of joints and numerous mechanical differences in construction. The humanoid android im building will have a very similar muscle,bone, tendon and ligament count and layout as a real person.
http://en.wikipedia.org/wiki/Human_skeleton
The key lessons learned from my dino build was that you can only cheat a little bit and the less cheating you do the better the final result. Once you have the basic techniques down then building a robot that uses 100s of tendons, ligaments etc becomes second nature. Its a bit like rigging a sailing ship, that said does get a bit hairy at times as the tension and motor torque is measured in 10s of kilograms, im not so much worried about losing a finger, im worried about losing an arm lol
*Note your Achilles tendon is designed to take a strain around half a ton !!! (do the math ) So ive been using Himalayan grade mountain rope and steel wire used for car vehicle recovery winches. Selected mountaineering rope is climate resistant and flexible but abrasion is iffy so it needs to be sheathed. To prevent sag over time use 3 to 6 times the minimum but but less tension per chord. Insertion points have to be carefully planned out and bedded in in order to prevent creep, backlash and related issues.
Some close up shots of the HRP android dancing.
http://www.youtube.com/watch?v=bLTdM8Y9nj8
OpenHRP3 (Open Architecture Human-centered Robotics Platform version 3)
http://www.openrtp.jp/openhrp3/en/about.html
Spent most of this weekend shopping for womens clothes, wigs and makeup etc....my cover story was shooting an indie film :-p ...that worked most of the time
Miss 'C' will be a tiny bit taller than HRP-4C , kind of like Goldilocks with attitude, lol
http://en.wikipedia.org/wiki/HRP-4C
She will also be 99% identical to a real human bodyform, HRP is close, close enough for most intents and purposes but she will never be able to move with the same grace as a real human due to the shortage of joints and numerous mechanical differences in construction. The humanoid android im building will have a very similar muscle,bone, tendon and ligament count and layout as a real person.
http://en.wikipedia.org/wiki/Human_skeleton
The key lessons learned from my dino build was that you can only cheat a little bit and the less cheating you do the better the final result. Once you have the basic techniques down then building a robot that uses 100s of tendons, ligaments etc becomes second nature. Its a bit like rigging a sailing ship, that said does get a bit hairy at times as the tension and motor torque is measured in 10s of kilograms, im not so much worried about losing a finger, im worried about losing an arm lol
*Note your Achilles tendon is designed to take a strain around half a ton !!! (do the math ) So ive been using Himalayan grade mountain rope and steel wire used for car vehicle recovery winches. Selected mountaineering rope is climate resistant and flexible but abrasion is iffy so it needs to be sheathed. To prevent sag over time use 3 to 6 times the minimum but but less tension per chord. Insertion points have to be carefully planned out and bedded in in order to prevent creep, backlash and related issues.
Re: HAL9000 thinking machine all the bells & whistles
Thor walking test
http://www.youtube.com/watch?v=aRvZQhSCUh4
Close up shots of the mechanics which are usually hidden from view in most bots.
http://www.youtube.com/watch?v=aRvZQhSCUh4
Close up shots of the mechanics which are usually hidden from view in most bots.
Re: HAL9000 thinking machine all the bells & whistles
Still chipping away at the mechanicals of my robot. Literally thousands of nuts and bolts and electrical connections. Each and every part designed ,prototyped and hand fabricated. The skull alone has over 1000 separate components. Slowly integrating the mechanics and electrics with the code as i go along. I at least have the satisfaction of knowing that it will be the first 'proper robot'
with a body and mind worthy of that claim.
*developed a new wiring loom using the finest gauge laminated motor wire. A close 'off the shelf' analogue to nerve fibers. The goal being 10,000 discrete sensors.
Some crazy robots you might have missed
Inside Adam Savage's Cave: Awesome Robot Spider! , probably the slickest robot made
http://www.youtube.com/watch?v=-vVblGlIMgw
Largest walking robot. Clear shots of the technical's, and a hint at the wider robo culture to come.
http://www.youtube.com/watch?v=PDJMZVcdNfc
Improvements in entertainment and hobby bots have been rather slow. This is more a reflection of the lack of 'off the shelf' plug n play components than a limitation of mankind's knowledge or technical ability. After all we put a man on a moon 50 odd years ago and just recently landed a truck sized robot on Mars in very spectacular fashion. The latest generation self stabilizing quadcoptors are pretty spooky in the flesh, here the RC hobby market has gained some traction and cost effective plug n play kit is available to any enthusiast.
Sooner or later someone is gonna drop a big bucket of cash into the hands of a group of artistically minded engineers and the klunkerbots of this past half century will instantly be forgotten. Such robots may be few and far between but they will be spectacular enough to be declared alive. If Asimo had a brain that matched its body you would have to treat it like some kind of person or being.
with a body and mind worthy of that claim.
*developed a new wiring loom using the finest gauge laminated motor wire. A close 'off the shelf' analogue to nerve fibers. The goal being 10,000 discrete sensors.
Some crazy robots you might have missed
Inside Adam Savage's Cave: Awesome Robot Spider! , probably the slickest robot made
http://www.youtube.com/watch?v=-vVblGlIMgw
Largest walking robot. Clear shots of the technical's, and a hint at the wider robo culture to come.
http://www.youtube.com/watch?v=PDJMZVcdNfc
Improvements in entertainment and hobby bots have been rather slow. This is more a reflection of the lack of 'off the shelf' plug n play components than a limitation of mankind's knowledge or technical ability. After all we put a man on a moon 50 odd years ago and just recently landed a truck sized robot on Mars in very spectacular fashion. The latest generation self stabilizing quadcoptors are pretty spooky in the flesh, here the RC hobby market has gained some traction and cost effective plug n play kit is available to any enthusiast.
Sooner or later someone is gonna drop a big bucket of cash into the hands of a group of artistically minded engineers and the klunkerbots of this past half century will instantly be forgotten. Such robots may be few and far between but they will be spectacular enough to be declared alive. If Asimo had a brain that matched its body you would have to treat it like some kind of person or being.
Re: HAL9000 thinking machine all the bells & whistles
So after all this time where are you one this awesome project, eh?
Re: HAL9000 thinking machine all the bells & whistles
TESLACOIL hit the singularity point, and was eliminated by his own AI?
Re: HAL9000 thinking machine all the bells & whistles
LMAO. I am pretty sure this was the same guy in another forum who said he was doing this awesome zombie movie project. Was hiring voice actors and everything. Maybe he made a Zombie HAL? LOLOLOLOL. Me, I think he's just delusional and finally got committed.