New Robonova...?
New Robonova...?
by tempusmaster
Sun Sep 30, 2007 9:00 am
I got excited when I saw the previous 'New Robonova' topic thinking that someone had news of a possible RN2 in the works. It turned out that the topic was about a new RN user - which was totally appropriate and actually quite interesting. But, it triggered a lot of though on my part about follow on products.
The Robonova has been around for a couple of years now, probably has the largest user base world wide among any of the humanoid kits, and has been blessed with lots of avid support from its owners.
Still, two years is a lifetime in this rapidly evolving new market. Lots of new designs like the KHR-1HV, i-SOBOT, Speecys, and RoboPhilo have appeared on the scene. The RN is still very competitive, but sooner or later a RN2 will be needed if Hitec wants to maintain its leadership.
So, what improvements would people like, or expect, to see in the next generation RN?
It goes without saying that we'd all like to see lower cost and higher performance. But, more specifically, what could Hitec do to improve the design to make it more compelling? What would a RN2 have to feature to make you consider buying one in addition to your RN1 or other existing robots? What would make a RN2 your next 'must have' purchase?
The Robonova has been around for a couple of years now, probably has the largest user base world wide among any of the humanoid kits, and has been blessed with lots of avid support from its owners.
Still, two years is a lifetime in this rapidly evolving new market. Lots of new designs like the KHR-1HV, i-SOBOT, Speecys, and RoboPhilo have appeared on the scene. The RN is still very competitive, but sooner or later a RN2 will be needed if Hitec wants to maintain its leadership.
So, what improvements would people like, or expect, to see in the next generation RN?
It goes without saying that we'd all like to see lower cost and higher performance. But, more specifically, what could Hitec do to improve the design to make it more compelling? What would a RN2 have to feature to make you consider buying one in addition to your RN1 or other existing robots? What would make a RN2 your next 'must have' purchase?
Latest robot news, information, reviews, hacks, photos, and videos - with special on-site coverage from Japan
http://www.robots-dreams.com
http://www.robots-dreams.com
by Pev
Sun Sep 30, 2007 9:29 am
Well an interesting question. My RN-1 still performs excellently and falutlessly after 18 months but technology has moved on so what would I like to see:
1. Hip rotation : This is a must especially after my experinces with the Bioloid Humanoid which has it.
2. Faster code execution : I find roboBasic can be a little slow
3. Extended IO capability (I have used all the AD ports now and also the ETX/ERX too) - maybe a better implimented I2C?
4. Extended support for the ETX/ERX prt - currently it can only seem to support single byte transfers
5. Extended battery Life maybe higher mAh Lipo's?
6. Some Sensors : Maybe include a basic sensor like a range finder as standard????? not sure on this one as not everyone would want it
7. Speach capability : Lets give the little guy a bit of a chance to express himself.
Anyway thats my first thoughts, and thanks for the Question Tempusmaster, certainly made me think about what I would want from a new RN-1.
Pev
1. Hip rotation : This is a must especially after my experinces with the Bioloid Humanoid which has it.
2. Faster code execution : I find roboBasic can be a little slow
3. Extended IO capability (I have used all the AD ports now and also the ETX/ERX too) - maybe a better implimented I2C?
4. Extended support for the ETX/ERX prt - currently it can only seem to support single byte transfers
5. Extended battery Life maybe higher mAh Lipo's?
6. Some Sensors : Maybe include a basic sensor like a range finder as standard????? not sure on this one as not everyone would want it
7. Speach capability : Lets give the little guy a bit of a chance to express himself.
Anyway thats my first thoughts, and thanks for the Question Tempusmaster, certainly made me think about what I would want from a new RN-1.
Pev
by Humanoido
Sun Sep 30, 2007 11:12 am
Tempusmaster, very good question, and excellent reply Pev!
The first thing that comes to mind is gyros. Above all, I want a robot that can walk in the most stable fashion.
Next, I want an accelerometer to be built in, so RN2 can walk up and down inclined surfaces, and know when and how to get up if it falls.
Speech is a must, especially for debug, and to verbally announce when the battery will need charging and when RN2 wants to nap.
I'd like more sensors, like a touch sensor, some vision, and especially two stereoscopic cameras that can sense distances, object shapes and colors. This would be used to pick up objects with realistic hands with thumbs and fingers. Face recognition software is a must!
The battery should go inside the COG center of gravity, and last longer. One hour performance is not enough.
Plus, RN2 will need more DOF. One more for the head so it can look around to see, without moving the entire body. Two more for the legs. Certainly more for the waist, torso, and hips, and some to turn the arm more realistically.
And last, more memory so we can work up some lengthy AI software and link it to motions and behaviors that can last all day. At night, RN2 can return to its charging bed and sleep all night to be rested enough to go at it again in the morning...
Of course, servos that can last all day without overheating.
I'd also buy an option to have it controlled over the internet. Hitec could have the knowledge base brain online and all the RN2s in the world could have great intelligence and world knowledge.
Tempusmaster, thanks for the great question!
humanoido
The first thing that comes to mind is gyros. Above all, I want a robot that can walk in the most stable fashion.
Next, I want an accelerometer to be built in, so RN2 can walk up and down inclined surfaces, and know when and how to get up if it falls.
Speech is a must, especially for debug, and to verbally announce when the battery will need charging and when RN2 wants to nap.
I'd like more sensors, like a touch sensor, some vision, and especially two stereoscopic cameras that can sense distances, object shapes and colors. This would be used to pick up objects with realistic hands with thumbs and fingers. Face recognition software is a must!
The battery should go inside the COG center of gravity, and last longer. One hour performance is not enough.
Plus, RN2 will need more DOF. One more for the head so it can look around to see, without moving the entire body. Two more for the legs. Certainly more for the waist, torso, and hips, and some to turn the arm more realistically.
And last, more memory so we can work up some lengthy AI software and link it to motions and behaviors that can last all day. At night, RN2 can return to its charging bed and sleep all night to be rested enough to go at it again in the morning...
Of course, servos that can last all day without overheating.
I'd also buy an option to have it controlled over the internet. Hitec could have the knowledge base brain online and all the RN2s in the world could have great intelligence and world knowledge.
Tempusmaster, thanks for the great question!
humanoido
by Gort
Sun Sep 30, 2007 11:55 am
Cool guestion.
1. All the servo's would be HSR-5498SG's
2. Would come with grippers
3. Double knee joints
4. Servo wires would come with wire protection already on them.
5. Lithium Polymer battery
6. Built in volt regulator
7. Low battery light that works.
8. Two gyro's
9. Tilt sensor
10. Increase the size and weight
11. A faster controller with more inputs and memory, an controller with dual cpus or quad cpus would be easy to do. Faster memory too.
12. A complier that works in Linux and Windows Vista
13. You could program it in Java or C##
14. A wireless G nic card.
15. A servo in the torso
16. Two ankle servo's so you can have left to right and front to back movement
17. Be able to rotate the grippers
18. Servo in the head and video camera
19. A wireless g controller with a built in Lcd screen link to the video camera
20. Two USB port connects to the controller
21. More room around the controller, for space for add ons
22. Carbon fiber parts
May add to this later, still thinking
1. All the servo's would be HSR-5498SG's
2. Would come with grippers
3. Double knee joints
4. Servo wires would come with wire protection already on them.
5. Lithium Polymer battery
6. Built in volt regulator
7. Low battery light that works.
8. Two gyro's
9. Tilt sensor
10. Increase the size and weight
11. A faster controller with more inputs and memory, an controller with dual cpus or quad cpus would be easy to do. Faster memory too.
12. A complier that works in Linux and Windows Vista
13. You could program it in Java or C##
14. A wireless G nic card.
15. A servo in the torso
16. Two ankle servo's so you can have left to right and front to back movement
17. Be able to rotate the grippers
18. Servo in the head and video camera
19. A wireless g controller with a built in Lcd screen link to the video camera
20. Two USB port connects to the controller
21. More room around the controller, for space for add ons
22. Carbon fiber parts
May add to this later, still thinking
by NovaOne
Sun Sep 30, 2007 12:23 pm
Hitec is a big enough company to go much further?
After seeing the video in limor's recent post http://robosavvy.com/forum/viewtopic.php?t=1777 and recently "playing" with an industrial robot arm, both these robots move in a fluid and precise manner, due to the use of inverse kinematics. I've still got a lot to learn when it come to the physics (theory and practical) of biped robots, but after hanging out at robosavvy for a few months I believe, properly integrating dynamic and kinematic information seems to be the next key step. Otherwise our RN's are little more than sequenced animatronic puppets.
I would like Hitec to:
1. Provide a development environment where you could edit simple kinematic chain model of RN-2, (when adding additional degrees of freedom), that could be compiled into kinematic transformation data for use in dynamic control.
2. Add a physics engine and a full editable 3D model, including mass and servo response information.....(maybe even virtual RN-2 programming environment?): some of you guys may have already done half the work for them; Fritzoid and Hivemind for example.
3. Download kinematic data from (1.) and dynamic data from (2.) to a gunstix (or similar) on board RN-2, to enable dynamic balance along with foot force sensors and a full IMU data. ie real time control using ZMP.
4. The gumstix could emulate the old MR-C3024 and run RoboBASIC if you wished, but a there should be a high level object oriented alternative which would enable advanced feedback control by beginners using pre-written objects, while being open enough for more advanced (robosavvy) users and universities to develop their ideas.
I realize all this is too much to ask from a toy manufacturer,
so if i "got real", I agree with Pev on all his points, but would also like to see arm/wrist rotation as well. It would be nice 
if RN-1 could be easily upgraded to RN-2 with a small kit costing less then say £300
Chris
After seeing the video in limor's recent post http://robosavvy.com/forum/viewtopic.php?t=1777 and recently "playing" with an industrial robot arm, both these robots move in a fluid and precise manner, due to the use of inverse kinematics. I've still got a lot to learn when it come to the physics (theory and practical) of biped robots, but after hanging out at robosavvy for a few months I believe, properly integrating dynamic and kinematic information seems to be the next key step. Otherwise our RN's are little more than sequenced animatronic puppets.
I would like Hitec to:
1. Provide a development environment where you could edit simple kinematic chain model of RN-2, (when adding additional degrees of freedom), that could be compiled into kinematic transformation data for use in dynamic control.
2. Add a physics engine and a full editable 3D model, including mass and servo response information.....(maybe even virtual RN-2 programming environment?): some of you guys may have already done half the work for them; Fritzoid and Hivemind for example.
3. Download kinematic data from (1.) and dynamic data from (2.) to a gunstix (or similar) on board RN-2, to enable dynamic balance along with foot force sensors and a full IMU data. ie real time control using ZMP.
4. The gumstix could emulate the old MR-C3024 and run RoboBASIC if you wished, but a there should be a high level object oriented alternative which would enable advanced feedback control by beginners using pre-written objects, while being open enough for more advanced (robosavvy) users and universities to develop their ideas.
I realize all this is too much to ask from a toy manufacturer,
so if i "got real", I agree with Pev on all his points, but would also like to see arm/wrist rotation as well. It would be nice if RN-1 could be easily upgraded to RN-2 with a small kit costing less then say £300
Chris
Last edited by NovaOne on Sun Sep 30, 2007 5:14 pm, edited 2 times in total.
by Gort
Sun Sep 30, 2007 2:18 pm
A 3-D modeling and programming environment sounds real cool. Just make sure you can map the servo settings of each individual robot not a generic servo map setting. I think Hitec should partner up with Sun Microsystems or Microsoft or Google to do the programming side.
You can do fluid moves with the RN, just do smaller adjustments to each servo for each step of a movement.
You can do fluid moves with the RN, just do smaller adjustments to each servo for each step of a movement.
by NovaOne
Sun Sep 30, 2007 3:12 pm
I agree about the programming team-up, most obvious would be MSRS.
Yeh, don't get me wrong fluid, movement is well within the ability of RN-1 hardware and software, if you spend the time to teach all the in between small steps.
What I was thinking of was more autonomous movement, whereby at the highest level/layer of your program, you could command RN-2 to walk or run by specifying a trajectory for the BCP (Biped Centre Point, ie centre of mass). The trajectory could be a simple linear vector, or more complex series of vectors to create a curve, and may be in what ever coordinate reference system you choose. A lower layer of code above the control loop would use it's kinematic and dynamic data of your RN and the surface level etc, to generate the actual move commands which could then be passed onto the control loop code which maintains dynamic balance using its ZMP algorithm.
Similarly you could program the hand/gripper (end effector) in tool frame coordinates with a TCP (Tool Centre Point) for example, like an industrial robot, lower level code would do the rest.
Of course you would program trajectory steps with a bluetooth teach pendant.
Chris
Lets hope someone from Hitec reads this thread.
Yeh, don't get me wrong fluid, movement is well within the ability of RN-1 hardware and software, if you spend the time to teach all the in between small steps.
What I was thinking of was more autonomous movement, whereby at the highest level/layer of your program, you could command RN-2 to walk or run by specifying a trajectory for the BCP (Biped Centre Point, ie centre of mass). The trajectory could be a simple linear vector, or more complex series of vectors to create a curve, and may be in what ever coordinate reference system you choose. A lower layer of code above the control loop would use it's kinematic and dynamic data of your RN and the surface level etc, to generate the actual move commands which could then be passed onto the control loop code which maintains dynamic balance using its ZMP algorithm.
Similarly you could program the hand/gripper (end effector) in tool frame coordinates with a TCP (Tool Centre Point) for example, like an industrial robot, lower level code would do the rest.
Of course you would program trajectory steps with a bluetooth teach pendant.
Chris
Lets hope someone from Hitec reads this thread.
Last edited by NovaOne on Mon Oct 08, 2007 7:31 pm, edited 1 time in total.
by TexasDave
Sun Sep 30, 2007 8:41 pm
I'd like to see quick release tabs to pop off the back cover so you can get at the control board instead of unscrewing four screws. I'd also like to leave the battery plugged in while popping off the back cover, instead of having to unplug it every time. It would have been a nice design to be able to remove the front cover plate without having to unscrew the control board.
by Mitchell
Sun Sep 30, 2007 9:54 pm
the first "new robonova" post was made by me... im still waiting for it, and dreaming about all the things i want it to do!!! for me, there is very little i would like to see Hitec add, cause then it would take away my fun of developing it.
the only thing i would like to see ( keep in mind i havent actually seen one in real life yet) is smaller/stronger servos and longer runtime. this would be so i could add more accessories of my own!!
the only thing i would like to see ( keep in mind i havent actually seen one in real life yet) is smaller/stronger servos and longer runtime. this would be so i could add more accessories of my own!!
by subpilot
Mon Oct 01, 2007 5:24 am
Some great ideas guys. I would really like to have a larger platform so I can add all the goodies that I would want. There's precious little space to work with in the current model. Of course that would drive the price up far beyond what mere mortals could afford to pay so it's just not going to happen.
Would at least like to see somebody make a super servo with some serious grunt so I could build a 3X size RN clone.
Would at least like to see somebody make a super servo with some serious grunt so I could build a 3X size RN clone.
