RoboSavvy Forum

Just a few thoughts...

2006-04-07T05:40:12+01:00

If you take a look at Most of the humanoids around today you will see there isnt a whole lot of room in them to put chain drives or other such things. You will also get reliabilty issues, servos tend to be pretty hardy and if they do break then they are easy to replace. You will find the reason that an a motor is not connect directly to a turn table is because it requires gearing down. Not many people really have much of a use for low torque high speed turntables. Servos are already geared down inside the casing, so they are actually ideally suited for connecting directly to the application, espically in the case of the robot servos that have the drive shaft coming out of either side of the casing. Unforunatly, if you have a great idea that has neever been done before its because a) Its just not practical or b) It has been done before you just dont know about it. If you do have a great idea that no one has thought of before, then patent it quickly lol. In regards to the other two questions, these have been done before, but these are really complex systems your talking about. Its often not done because its just to hard. If you want an idea of just how hard these things are then draw out an algarithum for these tasks. You'll soon see that its a big task. In saying that i would encourage you to not get discouraged. We must constantly push the envolope of what is done with hobby robotics, The technology you are talking about is cheap and readily available. There is nothing to stop you achiving any of these tasks. Dont do it because its easy, do it because its hard.

Statistics: Posted by Guest — Fri Apr 07, 2006 5:40 am

Re: Just a few thoughts...

2006-04-07T02:28:39+01:00

gsferrari wrote:
1] The servos are the load bearing structures in most of the robots today. Why arent there robots that rely on proper joints and axles that are simply rotated by the servos? Having the servo isolated from the joint by a "belt" or "chain" drive has certain advantages...this is just my opinion. Probably the same reason why an engine is not connected directly to the drive wheel...the motor of a turntable (in better turntables) is not directly connected to the platter...

The problem is that Hitec and Kondo and JR have been around for ages selling these ugly servo boxes. The Robo-One humanoids are a small new business opportunity extending the existing market for remote-controled cars, planes, sailboats etc. Since the planned sales of these robot kits are in the few thousand units per year, it is not worth investing in complex suspension systems, linear motors, continuous position reporting and feedback-control software.

2] Why isnt there a robot that has a basic controller to drive the motors - and a bluetooth or wireless tx/rx for transmitting telemetry and sensor output to a PC. This PC can handle the data, perform the AI computations and transmit the servo drive commands back to the controller.
The PC can also handle several instances of the AI program in the case of WiFi based on the IP address of the client robot...one PC can control several robots at the same time with different data-flows and robot actions.

Amen!
This is preciesly my philosophy. Take it a step further.. why not train a virtual robot in a virtual physics simulated 3D world? no bruises, faster timeline, automated re-enforcement learining.. take the "knowledge" that the virtual robot gained and apply it to the real robot.. Like in the movie Matrix (Simulacra)..
(BTW: I've started writing such a simulator called ezphysics.org (it's uploaded to sourceforge but the site is temporarily offline)).

3] Wireless controllers are at the point where they can pin-point clients on a map. If we can pin-point the robots on a map, the PC can then command the robots to perform "group-tactics"
Imagine 5 or 6 humanoids moving as a team -but- autonomously, to take out a target in a building. All we need is the floor map and access to the wireless controller and access point.

I'm hoping that we will soon see some development towards feedback-control by hacking the servos to provide continous feedback of position and power consumption to an alternative controler that supports wifi (like gumstix.com)

I cant wait for all this to become commercially available...The new Pirkus-II is supposedly 24" tall. Perfect for throwing in a wireless client device. I suppose this can also be done with the Robonova but the battery life is quite limited.

Don't hold your breath.

Statistics: Posted by limor — Fri Apr 07, 2006 2:28 am

Just a few thoughts...

2006-04-04T20:42:26+01:00

They werent kidding about the building process being a pain...the number of screws are crazy!!

Statistics: Posted by gsferrari — Tue Apr 04, 2006 8:42 pm

Re: Just a few thoughts...

2006-04-04T20:20:03+01:00

gsferrari wrote:
1] The servos are the load bearing structures in most of the robots today. Why arent there robots that rely on proper joints and axles that are simply rotated by the servos? Having the servo isolated from the joint by a "belt" or "chain" drive has certain advantages...this is just my opinion. Probably the same reason why an engine is not connected directly to the drive wheel...the motor of a turntable (in better turntables) is not directly connected to the platter...

Sucks to be at work with the message in my inbox that the Robonova-1 kit has just been delivered AARGH!!

The matter always is torque . The (cheap) motor technology inside the servo needs much RPM plus a gearbox to generate a significant torque. If this energy is transported through chains or belts, there would be too much backlash or jitter. Which will cause too much tolerance. One way out of this mess would be to use direct drives, i.e. brushless motors. These motors have the full torque, even at very slow speed, thus eliminating the gearbox.
The problem is, that with these motors the coils become too expensive, because they are very complex to wind. Therefore we have to use what is comercially available - unfortunately.
Have fun with your new arrival - I guess this will be a long night...

Statistics: Posted by Zeronar — Tue Apr 04, 2006 8:20 pm

Just a few thoughts...

2006-04-03T18:34:22+01:00

1] The servos are the load bearing structures in most of the robots today. Why arent there robots that rely on proper joints and axles that are simply rotated by the servos? Having the servo isolated from the joint by a "belt" or "chain" drive has certain advantages...this is just my opinion. Probably the same reason why an engine is not connected directly to the drive wheel...the motor of a turntable (in better turntables) is not directly connected to the platter...

2] Why isnt there a robot that has a basic controller to drive the motors - and a bluetooth or wireless tx/rx for transmitting telemetry and sensor output to a PC. This PC can handle the data, perform the AI computations and transmit the servo drive commands back to the controller.

The PC can also handle several instances of the AI program in the case of WiFi based on the IP address of the client robot...one PC can control several robots at the same time with different data-flows and robot actions.

3] Wireless controllers are at the point where they can pin-point clients on a map. If we can pin-point the robots on a map, the PC can then command the robots to perform "group-tactics"

Imagine 5 or 6 humanoids moving as a team -but- autonomously, to take out a target in a building. All we need is the floor map and access to the wireless controller and access point.

I cant wait for all this to become commercially available...The new Pirkus-II is supposedly 24" tall. Perfect for throwing in a wireless client device. I suppose this can also be done with the Robonova but the battery life is quite limited.

Any thoughts on these or other topics?

Sucks to be at work with the message in my inbox that the Robonova-1 kit has just been delivered AARGH!!

Statistics: Posted by gsferrari — Mon Apr 03, 2006 6:34 pm