An update of sorts.
Its been quite some time since I’ve been messing around with the Mindstorm kit. I’ve actually been slowly expanding my inventory with some of the new sensors from HiTechnic. In 2 months time, I would have had enough free time to actually come up with something (Exams start in 2 weeks and in 3 weeks time I would have gone home for the holidays).
Details about my next project:
I’m currently working on a school project. Its basically a game called Quoridor, programmed in Java. Its using a CLI, nothing fancy, but its my first foray in game AI. I’m using an Min-max algorithm with alpha beta node pruning. Maybe one day I’ll write up a basic short description of the algorithm. I would most probably be re-designing it after my exams as the current implementation is grossly inefficient and I already have some ideas of how to improve it. The re-design would most probably be coded in C++ and if its fast/efficient enough, I want to try to implement a NXT version, complete with a physical board and a robotic A.I. player.
UPDATE: Well, seems like the idea I had in the back of my head was very effective. Cut down the AI processing time by a lot. Works really fast now.
Might take some time though, the last time I programmed in C++ was like 4-5 years ago and back then the stuff I used was really basic. I would have to refresh my memory on C++. In the mean time, if you would like to read up more about the game Quoridor, here are some links:
http://en.wikipedia.org/wiki/Quoridor
Moving On…
Ah. Well well, the current design is still a failure. The gears were not locked properly because the body was not rigid enough. The body was flimsy due to the extreme weight and the lack of support structure. I did really like the drive train design though. It did work when implemented in a more rigid casing. Anyway the tank design will be on hold. I’m moving on to other projects first.
Things I’m thinking of doing:
1) Segway
2) Robotic arm with “personality” that grabs and plays with objects in its vicinity
Improvements and additions
Added a few things, improved on a few things.
There is now a front head light, controllable by the NXT “brain”. Using a Dual Longitudinal Subtractor drivetrain powered by two XL Power Functions motor with one Medium Power Functions motor powering the steering mechanism. Drivetrain connected to two Power Function Battery box. (Drivetrain is incomplete, waiting for differential gear box)
Decided against building a rotating turret. Too much work and I’ve too little mechanical knowledge to design such a thing.
Sensors:
HiTechnic Acceleration/Tilt Sensor
HiTechnic IRLink Sensor
HiTechnic EOPD Sensor
HiTechnic Compass Sensor
Weight : 1.8kg
Dimensions: 40cm (Length) x 20cm (Width) x 14cm (Height)
Time to start programming.
AT-UGV MK II
More parts have arrived. Now waiting for the final batch of gears including 2 differential gear box and then work can begin on the engine and the drive train.
I present to you, new and improved, AT-UGV MK II.
Did a test drive with the standard Mindstorm NXT servos and the tracks managed to climb a rather steep angle. About 60 degrees I would say. Not bad not bad, and mind you this is without all the fancy power gearing.
Rise of the Machines.
Building a robot can be quite a huge pain in the ass sometimes especially when you’re not good with fabricating metal parts and welding stuff together. Easy way out, use Lego Technic/Mindstorm NXT.
Anyway here is the preview of the robot I’m working on now. Its still in its infancy, I am just trying out different designs for the chassis. The following track system was conceived after 4 previous failed attempts. It has independent suspension for each wheel and will be powered by Power Functions motor connected to a IR receiver which will, in turn, be linked to the NXT using HiTechnic’s IR Link sensor.
Oh yes, this is only the proof-of-concept model. Going to make it bigger so I have more space to mount more stuff.
