[Prev][Next][Index][Thread]

RE: mouse performance



Hi all
 
Good to see Duncan kicking up some dust again!
 
As Pete suggests, we are planning an event in late November. This will be an informal competition ... held in the TIC itself, not in thinktank theatre. I am currently sorting a date, and we are hoping to have one or two other events such as drag races and time trials, as well as the wall follower and maze solving competitions.
We may have a seminar running at the same time, targeted at the much more general theme of pervasive/ubiquitous computing - that depends on whether the building work currently underway is completed on time, as one of our labs has been pulled apart over the summer. Whatever, we will hold the mouse competition - but we need a name for it! We have Micromouse, Minos, and ?????
As it is the last time the mice will be seen until Easter, perhaps it should be a (hiber)national event? 
 
Some ground rules need to be set - some VERY SIMPLE ground rules - so that you good people out there can start preparing. We will be having our August meeting of the Midlands Micromouse club on Wednesday August 3rd, so suggestions please. I will ensure that we have a firm date before the end of that week.
 
I have used the rectangle thing as a test track for student wall-followers. I think a known track such as this give us a benchmark figure for mice. I also thought of having the maze re-arranged to form 4 lanes by 32 cells for drag racing - no parachutes allowed.
 
Tony
 
 

	-----Original Message----- 
	From: owner-micromouse@cs.rhul.ac.uk on behalf of Peter Harrison 
	Sent: Sun 7/24/2005 5:45 PM 
	To: micromouse@cs.rhul.ac.uk 
	Cc: 
	Subject: mouse performance
	
	

	I don't think there is a single source of mouse performance data
	although there are a couple with records of precious competitions. These
	times are, of course heavily dependant upon the actual maze.
	
	I have found data for a couple of mice on the net (you may need to mess
	with your font settings to see this neatly) :
	
	Name            Ning 2          Min2            Min3            MM3
	owner           Ng Beng Kiat    Ng Beng Kiat    Ng Beng Kiat    Itani
	length  mm      122             117             113             130
	width   mm      80              75              74              60
	height  mm      70              50              41              52
	mass    g       713             332             290             350
	speed   m/s     2.5             2               2.1             2.6
	accn    m/s/s   2.5             2.5             2.8             3.2
	
	For what it is worth, Maximus (you remember - knocks down walls and
	crashed a lot) was set for an acceleration 2.8 m/s/s and a top speed of
	2 m/s.
	
	It is clear from the table above that a mouse that can handle
	consistently at 3 and 3 would be in with a good chance against most
	competition.
	
	All the above mice can accelerate much harder and travel much faster. I
	believe these are the actual running settings used.
	
	If anyone else would like to share their performance data, I am sure we
	would all be interested. It is good to know what you have to achieve to
	be in with a chance. Now, controlling all that power....
	
	The only common test I have come across on the Japanese sites is a full
	circuit around the outside of the maze. The simplest way to set this up
	would be to time the mouse between subsequent passes through the start
	gate, having removed the east wall from the start cell. Mice would then
	just need to be told to do a simple set of moves. I have asked Tony if
	we can have such an event at the proposed semi-formal  autumn/winter
	event at TIC. We are looking at the end of November I think. More
	details from Tony if we can go ahead with this. I have a video somewhere
	of a stepper driven mouse doing this in under 10 seconds and I think 6
	seconds would be a fair goal for a mouse accelerating at 3m/s/s.
	
	As for how useful it is, well the ability to steer at speed is fairly
	critical and it is a simple, easy to set up task. Any well-defined,
	repeatable task would be interesting as a year-on-year comparison of
	performance. I have been unable to find published data for this although
	I am sure it exists somewhere. It may also be a good crowd-pleaser at a
	competition.
	
	Pete Harrison
	

<<winmail.dat>>