While I'm still building my first robot, I'm already thinking about choosing a platform for my next project. I'm currently using a simple two-wheel differential drive base that is not ideal for my purposes. First of all, it is very bad at moving on anything else besides hard floor and cannot climb over even the smallest thresholds. Therefore I'm looking at a tracked or 4WD kits. Second, it's getting crowded - while I'm using an Arduino to control the servos and sensors, I also have a Chumby One that is going to be the main computer onboard.
I've looked at what RobotShop has to offer, and currently there are three kits on my shortlist:
I want to be able to mount an Arduino, my Chumby One, a small camera, sensors and some sort of a gripper or an arm. Therefore the DFRobotShop Rover might be just too small, even with the optional extra deck. But it's cheap and therefore tempting. The problem with Lynxmotion 4WD1 is that I don't think it will be easy to install rotary encoders on the motor shafts, or am I wrong? So the Traxster II is currently my top choice.
I wrote a blog entry on this subject, there are links, images, information on the three alternatives I mentioned, and more complete information on my requirements. Please have a look:
Any comments or suggestions would be most welcome. Does anyone here have experience with the three kits I mentioned? Any other alternatives?
zoomkat:
[What is the function of the rotary encoders? The bots you are looking at by design have a lot of slippage on their operating surface.
I was planning to implement closed-loop control, to enable the robot to drive as straight as possible etc. Therefore I'd like to measure the angular velocities of the wheels. I didn't realize that slippage would pose such a major problem.
Wouldn't encoders at least help in detecting slippage and compensating for it? Combined with other sensors like an electronic compass or an accelerometer maybe?
zoomkat:
[What is the function of the rotary encoders? The bots you are looking at by design have a lot of slippage on their operating surface.
I was planning to implement closed-loop control, to enable the robot to drive as straight as possible etc. Therefore I'd like to measure the angular velocities of the wheels. I didn't realize that slippage would pose such a major problem.
Wouldn't encoders at least help in detecting slippage and compensating for it? Combined with other sensors like an electronic compass or an accelerometer maybe?
Thanks for your comment!
I'm not a robot designer or even user, however I never quite understood why they design their robots such that their speed control is also often the steering control. Have we not learned anything from a century of driving cars, my hands control the steering and my right foot the speed. Think they place a much harder burden on the software trying to intergrate these different control objectives.
retrolefty:
I'm not a robot designer or even user, however I never quite understood why they design their robots such that their speed control is also often the steering control. Have we not learned anything from a century of driving cars, my hands control the steering and my right foot the speed. Think they place a much harder burden on the software trying to intergrate these different control objectives.
Any thoughts on that?
I'm not an expert, but my gut feeling is that separating speed control and steering wouldn't make the overall problem any easier. The robot would still have to perform the same basic tasks, i.e. maintain the desired speed and direction. Now it would just have to control more actuators and perhaps monitor more sensors, which could even make the control algorithm more complicated. Moreover, differential steering has the advantage that it is easy to make the robot rotate around its axis without moving forward.
I've been working on a Rover 5 platform that I go from sparkfun.com. There are a few different variations, but the '5' has 4 motors, each with it's own quadrature encoder, and is tracked. Since it does have independent 4-wheel drive, you could use mecanum wheels for an extra level of maneuverability.
Robosavvy.com makes a motor controller/encoder board that looks like it works very well with this platform, but I haven't tried it yet.
deravor:
I've been working on a Rover 5 platform that I go from sparkfun.com. There are a few different variations, but the '5' has 4 motors, each with it's own quadrature encoder, and is tracked. Since it does have independent 4-wheel drive, you could use mecanum wheels for an extra level of maneuverability.
Thanks for the suggestion! I like that platform, but I'm a bit concerned that it would be too small (like the DFRobotShop Rover). The Chumby takes up a lot of real estate.
But I could perhaps mount two decks on top of the robot, a bottom deck for the Arduino, motor controller etc. and a top deck for the Chumby and a camera.
I've been stuck in the same situation. The robotics hobby has a absolutely huge lack of tracked chassis to choose from right now. You have your selection which don't meet any of my needs really. So it boils down to buying old/new RC tanks and ripping the guts out. The Tyco Fast Traxx line look like a good tracked platform if you can find any. The RAD 2 robot looks pretty good as well. However, even in the RC toy line, not many people make tracked anything anymore.
This has been the most frustrating thing about making a tracked robot for me. The chassis is impossible to find unless you want to spend upwards of $300 for something that goes above and beyond a basement hobby.
Philthy:
I've been stuck in the same situation. The robotics hobby has a absolutely huge lack of tracked chassis to choose from right now. You have your selection which don't meet any of my needs really. So it boils down to buying old/new RC tanks and ripping the guts out. The Tyco Fast Traxx line look like a good tracked platform if you can find any. The RAD 2 robot looks pretty good as well. However, even in the RC toy line, not many people make tracked anything anymore.
This has been the most frustrating thing about making a tracked robot for me. The chassis is impossible to find unless you want to spend upwards of $300 for something that goes above and beyond a basement hobby.
Have you thought about building your own tracked chassis?:
deravor:
I've been working on a Rover 5 platform that I go from sparkfun.com. There are a few different variations, but the '5' has 4 motors, each with it's own quadrature encoder, and is tracked. Since it does have independent 4-wheel drive, you could use mecanum wheels for an extra level of maneuverability.
Thanks for the suggestion! I like that platform, but I'm a bit concerned that it would be too small (like the DFRobotShop Rover). The Chumby takes up a lot of real estate.
But I could perhaps mount two decks on top of the robot, a bottom deck for the Arduino, motor controller etc. and a top deck for the Chumby and a camera.
Hey mate,
I've got the Rover 5 and I'm pretty happy with it, here's my setup:
If you put a larger panel on the base you'd have a fair bit of space to play with.
Have you looked into Rc rock crawlers? They can drive over extreme terrain and the suspension style allows for a good amount of space in the chassis to mount electronics.