Microb Technology/2006
De Wikidroids
Sommaire |
NEWSFLASH!
The French Robotic Championship 2006 is over. Microb Technology performed well, being present in the final act of the competition and being ranked 10-th at the end of the qualification rounds! (171 teams were present) We are, of course, hyermotivated to perform even better in 2007!!
Microb Technology
Microb Technology is a robotic team. Some members of this team are members of the Droids Corporation, so that's why the wiki is hosted here.
Commercial booklet :
Commercial booklet, inside (.pdf)
Commercial booklet outside (.pdf)
Members
The members of Microb Technology are :
- Vincent MESUREUR : President of the association, mechanic of the robot and commercial partnerships..
- Laurent CORNAILLE : Electronic specialist. Design the motherboard and the powerboard.
- Florentin DEMETRESCU : Electronic and computer sciences artist, design of the camera board and development of the low level image acquisition code for dsp.
- Fabrice DESCLAUX : Computer sciences r0x0r, development of the image processing code (multiplatform, to be implemented on dsp).
- Olivier MATZ : Computer sciences, development of the robot main code (maintener of Aversive)
Occasinnal's Members:
- Christoph(e) RIEHL : Professionnal Sabotager
- Thibault VINCENT : Commercial partnerships.
Rules 2006
The 2006's rules are available (in english).
They are available in french too.
Playground
Juste some picture of the realisation of our playground. Follow this this link.
Strategy
The first thing we did before building anything, was to think about our strategy.
We don't have a lot of time:
We have 14 holes of our color, so this mean that we can only spend 90/14 = 6.42 second for getting balls, and put them in the holes.
- we can win time be detecting the balls with a camera
- we should optimize trajectory and ball management in the robot
Mechanics
Positioning System :
We think that a powerful positioning system will be really mandatory for dropping balls in holes. Odometry could be a good choice. Last year we had 4 wheels on the robot (2 motorised ones, and 2 free wheels that only measure distance). The problem with free wheels this year is that there are holes on the area !!
- The chosen solution will be a software one : avoid moving on holes, and try to detect bad encoder values.
Handling balls :
The harvesting solution appears to be easier than expected. Hoover balls is fast.
What we need to handle balls :
This is just a litle sumary of all we need for our handling system :
- 2 motors for propultion wheels.
- 2 controled harnesses (On/Off).
- 4 ServoMotors (catching, selecting and droping balls).
- 1 motors for controling the forward roller.
Some pictures :
For pictures of mechanic just follow this link.
Electronics
Our choices for 2006 :
Last years we had choose to build a modular electronic. Functions were grouped on individuals smalls cards. This had the following advantage :
- Used space is proportionnal to the numbers of function used in the robot. If you have few functions you have a few cards to build.
- If you have identical functions in your robot, you dont need to developpe a biggest card.
- If you find a bug (an electric one) on a card you can correct it on the others.
- It is easier to find a place for it in the robot, even if you dont know how mechanics is.
This years we did the choice of a compact electronic "all in one" for the following reasons :
- Now we will have to drive up to six motors. So this means if we do it modular we need 3 Powerboard, 3 Motherboards, 1 or 2 Power_suply_Board, and 2 cards for Sensors.
- A legion of cables to link the cards. It take us about 16 hours to build all the cable in 2004, and we had only 4 motors. I don't speak about the time spend for the sensors.
- On the power board, a place is reserved for a big cooler which is shared by 4 LM18200T. With modular board you have some kind of mechanical restrictions.
More details
For more details about our electronic system, have a look to this link.
Vision system
Our vision system is based on an evaluation board from Texas Instruments which integrates a DSP processor (TMS320C6416T) running at 1GHz and having a maximum processing power (theorical) of 8 GIPS (Giga Instructions per Second). With this processing power we hope to be able to run our image recognition algorithms in real time (24 frames per second).
On the image acquisition side an interface board is in developpment. It features a video decoder circuit also from Texas Instruments (TVP5146) and an FPGA module (manufactured by Dallas Logic) integrating a small FPGA from Altera (Cyclone I - EP1C3). Also a small memory (512 kB of SRAM) is present on the camera board. It serves as a buffer for the acquired video frames, until the DSP commes to read it. The FPGA manages all the operations on the board and serves as an interface for the DSP. The camera board also integrates a rs232 interface which will be used to communicate with the main board of the robot.
Our vision system can work whith any camera which generates a composite video output. Actually, we are using a very cheap one, but the image quality is still corect (see image section for some examples, or here).
Main program
Planning
- october : Integrate the following modules : list, uart, utils, wait. Start the motor control design document.
- november : Integrate scheduler, pwm, counter_microb. Start the coding of motor control system.
- december : Finish the motor control system.
- january : Tests and fixes on motor control system with the real robot. Start trajectory and position manager.
- february : Test trajectory and position modules.
- march : Conception of strategies, code the program for other sensors (camera, ...) and other systems (TBD)
- april : Strategies
- may : Tests, tests, tests.
It's very diffucult to respect milestones in the project, so we created a page with our Planning (in french).
Documentation
Microb Technology uses/develop the Aversive project for AVR microcontrolers. Here you can find a documentation (in french) about our control system.
You can find our complete code here :
- Aversive framework
- Microb2006 project
- Documentation of aversive
- Communication protocols (between motherboard, camera, PC)
The robot 2006
Partners
We want to thank our patners for their help : This page describe in details who are our partners.
Budget
As one can notice it, the robot is expensive. Fortunately, our sponsors help us by providing us their products, which decreases the financial investment of the members of the team.
