I got my robot last month, and the backlash frustrated me a lot. Does anybody test the backlash of dobot and how is it? Mine is around 0.5-0.6 degrees. What's more, if it has the encoder it will also be better to do the compensation, but it only has the IMU modules:(
Is there any other robot with better performance?
I think my dream robot should have: 1.backlash under 0.1 degrees; 2.encoder in each joint; 3.controller based on RAMPS; 4.payload around 500g; 5.small enough to put on the desk; 6.price less than $800.
Unfortunately, currently 1&5&6 together is not likely to happen. The problem is the reducer. If you want it cheap it will either have backlash (with regular gears) or will be bulky (if your reducer is based on timing belts).
RAMPS was designed for Arduino Mega, and stands for "RepRap Arduino Mega Pololu Shield" reprap.org/wiki/Arduino_Mega_Pololu_Shield Arduino Mega is hosting quite a slow 8-bit 16MHz microcontroller and it's quite hard to fit lots of stuff there. Remember how dobot went from 256000 serial rate down to 9600 because of stability? It simply wasn't capable of not missing data on serial interface at that rate and still drive the arm. So I'd suggest to keep your mind open for other things that may still do the thing, like this www.sparkfun.com/products/13907 (dual-core 240Mhz 32-bit). You can still do a lot of stuff "adruino-style" with it (e.g. learn.sparkfun.com/tutorials/esp32-thing-hookup-guide?_ga=1.36541024.1378430733.1483237465#arduino-example-blink) and use many arduino libs (as long as they are pure arduino-api without any assm includes and direct hardware access), yet it has many other powerful features. You'd, of course, need to do some soldering, but not much - just wire motor drivers and power them.
Yeah, you are right. Ramps with Arduino Mega is not powerful enough to do both the communication and controlling. But it's still best choice for the beginners to build up a open source 4-axis robot. I should have a try with the ESP32 as you said. Thanks Max! The reducer is a big problem for the robot. But considering the usage like 3d printing, laser engraving, the precision we need is much lower than the industry level, there still might be some chance for us to improve the performance. Maybe in the future, There will be something we want.
I must say there is a price gap between professional manipulators (starting at lets say 15,000$) and these "toys" like dobot costing around few hundreds dollars. I didn't see robots with the price between these margins (which could at least have some of your requirements satisfied).