NXT Robo-One?
I've been watching some of the Robo-One competitions and am really amazed at what these little guys can do. I realize these bots are slightly more complicated in their programming (balancing, for example, requires some finesse) and have more servo motors than the basic retail kit, but I can't stop wondering what an NXT version would look like.
Given the size and shape of the NXT servo motors, I think it's safe to say that a NXT Robo-One competitor would be slightly larger in size. I haven't dug too deep into the rules, but I imagine there are maximum and minimum height/weight requirements (if anyone knows more, please comment).
A bot like the Robonova-1 has a total of 16 servos. Given that we will be able to buy individual servos from LEGO for under $20US, I don't think it's completely impossible for us to expect some individuals to purchase 13 or more servos and build a "NXT-One."
One of the biggest costs for buying and/or building a custom Robo-One competitor seems to be the machining costs involved in building the framework to hold all those servos - add in the cost of the software and the servos and you've got a pricey little bot. The ready-to-assemble kits float around $750 to $2000 or more (Robonova-1, Lynxmotion, Kondo).
Now, let's tally up the costs for the NXT-One:
Basic Retail Kit: $250US
13 additional servos: $260US
Software: $0
There will be some other costs not listed (such as extra cables), but the biggest expense are the motors. The framework would be built from the basic Technic beams and misc. parts. If the NXT-G software can't handle the programming requirements, there are plenty of free options that are available now or coming soon. But I think it's very reasonable to see a few bi-ped bots designed using the NXT system showing up soon...
Now, are there potential problems? Sure.
The cabling for the NXT sensors and motors are thick and not so flexible. Easy to fix? Absolutely - I'm sure users will figure out how to create more flexible cables, not to mention third-parties offering different lengths and thicknesses for the cables. LEGO may have options in the planning as well...
Size is definitely an issue. If you figure 1 servo for foot, ankle, calf, thigh, waist, shoulder, bicep, and forearm (x2 for 16 servos), you've got one very TALL bot. Take a look at the example Robo-Ones I've linked to and you'll see that most of these servos are small, square, and fairly compact. So we've got the issue of the NXT motor size and shape to deal with. But LEGO builders are a creative bunch, so I don't anticipate this being a problem for too long.
Balance - I've seen NXT bots now that balance okay on two legs... but the steps taken are small, and some could be considered sliding rather than true walking. This will most likely be a software issue - how best to shift the weight/center-of-gravity of the NXT-One to keep it from falling over.
Bulkiness - It'll be interesting to see how few parts can be used to build a solid framework for keeping all the servos linked while at the same time reducing bulk and keeping the bot looking "human" in shape.
Other issues? Please comment...
Jim
Given the size and shape of the NXT servo motors, I think it's safe to say that a NXT Robo-One competitor would be slightly larger in size. I haven't dug too deep into the rules, but I imagine there are maximum and minimum height/weight requirements (if anyone knows more, please comment).
A bot like the Robonova-1 has a total of 16 servos. Given that we will be able to buy individual servos from LEGO for under $20US, I don't think it's completely impossible for us to expect some individuals to purchase 13 or more servos and build a "NXT-One."
One of the biggest costs for buying and/or building a custom Robo-One competitor seems to be the machining costs involved in building the framework to hold all those servos - add in the cost of the software and the servos and you've got a pricey little bot. The ready-to-assemble kits float around $750 to $2000 or more (Robonova-1, Lynxmotion, Kondo).
Now, let's tally up the costs for the NXT-One:
Basic Retail Kit: $250US
13 additional servos: $260US
Software: $0
There will be some other costs not listed (such as extra cables), but the biggest expense are the motors. The framework would be built from the basic Technic beams and misc. parts. If the NXT-G software can't handle the programming requirements, there are plenty of free options that are available now or coming soon. But I think it's very reasonable to see a few bi-ped bots designed using the NXT system showing up soon...
Now, are there potential problems? Sure.
The cabling for the NXT sensors and motors are thick and not so flexible. Easy to fix? Absolutely - I'm sure users will figure out how to create more flexible cables, not to mention third-parties offering different lengths and thicknesses for the cables. LEGO may have options in the planning as well...
Size is definitely an issue. If you figure 1 servo for foot, ankle, calf, thigh, waist, shoulder, bicep, and forearm (x2 for 16 servos), you've got one very TALL bot. Take a look at the example Robo-Ones I've linked to and you'll see that most of these servos are small, square, and fairly compact. So we've got the issue of the NXT motor size and shape to deal with. But LEGO builders are a creative bunch, so I don't anticipate this being a problem for too long.
Balance - I've seen NXT bots now that balance okay on two legs... but the steps taken are small, and some could be considered sliding rather than true walking. This will most likely be a software issue - how best to shift the weight/center-of-gravity of the NXT-One to keep it from falling over.
Bulkiness - It'll be interesting to see how few parts can be used to build a solid framework for keeping all the servos linked while at the same time reducing bulk and keeping the bot looking "human" in shape.
Other issues? Please comment...
Jim
Comments
Someone also talked about having 4 motors for symmetric reasons; and the fact TLG actually sells motors separately whereas you got enough in the box is probably a clue in that direction...
But still I was under the impression you could only connect one motor per output. Am I wrong? And if you can connect more, is this easy or is some trickery (hardware/software) required?
Now, after using the product, it raises even more questions. I don't know if a multiplexor would work with the NXT-G software included with the retail kit, mainly because the programming blocks require you to specify motors A, B, or C when issuing commands to the motors. Unless the software is modified to allow for a higher number of motors, I don't see how the MPlexor would allow you to communicate to any more than 3 motors.
That said, it'll probably come down to a software alternative that has the ability to control more than 3 motors.
Also, the multiplexor will allow for multiple sensors (beyond the standard 4 ports on the brick) but again, this raises the question of how the NXT-G software can control say Light Sensor #2 versus Light Sensor #1... I just don't have enough information yet on how something like this would incorporate into the existing software (and it might not).
Thanks for bringing this up... excellent point.
Jim
The article is at this link:
http://www.theglobeandmail.com/servlet/story/LAC.20060621.IBLEGO21/TPStory/Business
William
http://www.theglobeandmail.com/servlet/story/LAC.
20060621.IBLEGO21/TPStory/Business
--
Brian Davis
I've been looking into them for a while, and it seems as though the framework and software are basically free. For the Robonova, when you add up the retail price of the specialized servos that they use, the controller, the power supply, and the remote, you are well beyond the $1000 or so that they are charging for the kit.
I look forward to seeing what kind of Robo-One type humanoid can be built with this Lego kit. If a fully articulated humanoid can be built with such cheap servos, I might actually be able to get one myself.
Keep up the good work!
You're absolutely right... most of these kits do include the software. But if you were to buy the servos separately and add in the software and mounting hardware, I imagine the price would be even higher than the ones I specified. Most of the Robo-Ones being built for competition are one-of-a-kind designs - there just aren't that many Robo-One kits right now.
And while I'm back on the subject of Robo-Ones, it did occur to me that 5 NXT kits would be all it takes to build one of these - the 5 bricks would communicate via Bluetooth to coordinate movement.
And the best thing about a NXT-One? Sensors! Most Robo-Ones don't have sensors, just servos for all the fancy movements. Imagine a NXT-One biped that also has the functionality of up to 20 sensors!
Jim
On the flip side, I notice that one of the (eventual) optional accessories will be sensors and perhaps a BT module for the Robo-one control board. Depending on the format, with BT a Robo-one might make a nice slave device for a NXT... I'm picturing a NXT on a cart, being pulled by a harnessed Robo-one... :-)
--
Brian Davis
Jim
The biggest single issue that immediately comes to mind is the servos, torque, and load bearing. With a ROBO-ONE the servos in the feet bear the entire weight of the robot and have to be able to move it. It's not unusual to strip gears or burn servo motors, which is why the kits use the beefier, more expensive servos.
I'm not saying that it can't be done - it probably can, and I'd love to see it happen. It's highly likely that Jin Sato already working on one since in addition to his Lego creations he's responsible for the Pirkus ROBO-ONE robot series and enters the competitions here regularly.
As for the ROBO-ONEs not having sensors, that's a common misconception. The basic kits don't come with sensors, but there are lots of options available, at least here in Japan. A lot of the top ROBO-ONEs have multiple sensors and are capable of very basic autonomous behaviors. Even the KHR-2HV kit that was introduced a few weeks ago uses the RCB-3J controller that has several digital and analog sensor inputs and a graphical programming application.
Some of the larger robo one bots double up the servos in the joints to (a) reduce the arcs from 180 degrees to 60 degrees for each servo (this increases speed), or side by side to increase strength.
I'm it would be possible to double up NXT motors.
I would think 4 motors per joint, double strength and 1/2 the arcs.
Also all 4 motors could be connected to act on one signal, since 1/2 would be moving in reverse that would be mean reversing the command signal and in regards to the pairs, just send the same signal to both motors.
Regards
Michael