Comment Re:Interesting Simulation/Robotics project (Score 1) 108
To the guy that commented on it being a "pre-school project"....
I think you misunderstand the background.
There is no robotics course here, and no background in robotics at all. We have no lecturers in robotics, and myself and this other guy are the only people with any experience at all - and we're just beginners. Most of his project and a fair bit of my time has been spent trying to build up a knowledge of the area.
The project made up either 15 or 30 credits of a 180 credit course over 7-8 months. (I'm not sure). So the guy didn't spend a year on it, more like 200 hours for everything (see below).
The point of the project wasn't just to demonstrate those behaviours mentioned, but instead to;
1) Practice the full software lifecycle (research, design, implementation, testing, and everyone' favorite, documentation... typically 80 pages of detailed A4 text in one of the reports alone, never mind javadoc etc.).
2) Develop a piece of software that would make it easier for people to develop whatever sorts of behaviour they want, and make it easy for people to plug in their own behaviour, on any platform, and make it easy for people to extend his simulator, or reuse it , or whatever.
3) Make it so that people could use the same bit of source code for working on-board, , off-board , or working in simulation.
4) Then, develop some plugins to show what he'd done.
I'm sure someone could have done the project better had they been spoon-fed the material by someone experienced in the area, but the guy doing this had to teach himself neural nets, reinforcement learning, robotics, etc. etc as well as get familiar with java, lejos, swing and so on, all within that 200 hours.
At the start of this, we didn't even know what robotics platforms were out there, we had to spend time checking them all out, and trying to scrape together some money once we decided a mindstorm would be fun!
Anyway, you have a fair point in that it's not the best project in the world, but it's fun, and it makes it easier to rattle off controllers in for on-board, off-board processing , and for simulation.
I think it was a good 200 hour project, given what the project tried (and succeeded) in doing, and the resources available. It also managed something that I don't think any other program does- a simulator that allows you to run the same java code in simulation or on the RCX without editing / recompiling, on any platform. (linux/windows)
If you still feel the same now that you know the background, then that's your choice, but have a play with the program and see if you like what it can do.
I think you misunderstand the background.
There is no robotics course here, and no background in robotics at all. We have no lecturers in robotics, and myself and this other guy are the only people with any experience at all - and we're just beginners. Most of his project and a fair bit of my time has been spent trying to build up a knowledge of the area.
The project made up either 15 or 30 credits of a 180 credit course over 7-8 months. (I'm not sure). So the guy didn't spend a year on it, more like 200 hours for everything (see below).
The point of the project wasn't just to demonstrate those behaviours mentioned, but instead to;
1) Practice the full software lifecycle (research, design, implementation, testing, and everyone' favorite, documentation... typically 80 pages of detailed A4 text in one of the reports alone, never mind javadoc etc.).
2) Develop a piece of software that would make it easier for people to develop whatever sorts of behaviour they want, and make it easy for people to plug in their own behaviour, on any platform, and make it easy for people to extend his simulator, or reuse it , or whatever.
3) Make it so that people could use the same bit of source code for working on-board, , off-board , or working in simulation.
4) Then, develop some plugins to show what he'd done.
I'm sure someone could have done the project better had they been spoon-fed the material by someone experienced in the area, but the guy doing this had to teach himself neural nets, reinforcement learning, robotics, etc. etc as well as get familiar with java, lejos, swing and so on, all within that 200 hours.
At the start of this, we didn't even know what robotics platforms were out there, we had to spend time checking them all out, and trying to scrape together some money once we decided a mindstorm would be fun!
Anyway, you have a fair point in that it's not the best project in the world, but it's fun, and it makes it easier to rattle off controllers in for on-board, off-board processing , and for simulation.
I think it was a good 200 hour project, given what the project tried (and succeeded) in doing, and the resources available. It also managed something that I don't think any other program does- a simulator that allows you to run the same java code in simulation or on the RCX without editing / recompiling, on any platform. (linux/windows)
If you still feel the same now that you know the background, then that's your choice, but have a play with the program and see if you like what it can do.
