Darwinbots Forum
Code center => Bugs and fixes => Bug reports => Topic started by: Elite on April 14, 2006, 06:21:59 AM
-
Firstly, well done to Eric - you've done a great job here
There are only three things that bug me in this new release:- Collision detection
- *.thisgene
- Ties
Collision detection doesn't seem to work. Even when I line up veggies in a wall arrangement, Minimalbot (attached) just seems to go straight through. When I do manage to get Minimalbot to 'hit' a veggie, Minimalbot seems to take damage because of it.
Viruses that use *.thisgene to propagate don't show up and have a vastly decreased range: compare a virus that uses *.thisgene to a bot that places a normal number in .mkvirus
Ties are acting slightly dodgy, but I can live with this for the moment
-
You're comming from 2.3X right?
If I remember right I fixed a virus range bug in the code in 2.4 a while ago. The virus problem might be related to that.
Collision response has been quite shaky in 2.4. It's one of its problems. I don't know about the taking damage, but I routinely see bots seem to just pass through each other. Its because the collision response isn't strong comapared to their momentums. This has been fixed in the C++ version, I'm happy to say, so I don't know how much work should be spent fixing this in 2.4.
Back to the damage comment, maybe we should damage bots that collide with something too fast. That would encourage bots to watch their speed while still letting them use high speeds to escape from potentially fatal situations.
As to ties, is it tie physics that doesn't work or another aspect of ties? Ties seem to work for me better in physics than in 2.3X. Try making a bot that only reproduces once and then ties to the offspring. Drag and throw one of the bots, and you should see some rather realistic motions.
-
I wonder about the realism with putting collision damage into this (here I go again ). With such light weight organisms living in a liquid fluid I don't think they should take any damage at all. Maybe if they bounce into each other at very high speed it whould make sense. I think light organisms require very high force to get damaged. A fly flying into the window doesn't seem to get any damage for example.
-
Flies tend not to get hurt because of their thick exoskeleton. Though I have managed to kill flies with a fly swatter in mid air by hitting them hard enough. I dunno, this should probably be in the suggestions forum I think.
-
Tie physics are acting OK, except for the massive change in springyness, but my Symbioticus (attached) doesn't seem to be unfixing it's veg in 2.4
I just noticed that viruses that use *.thisgene are noticably more flickery. They just don't show up very well - it's kinda weird. Try it
As for collision detection, well, it affects the survivability of minimalbot, since minimalbot depends on 'sticking' to veggies to feed (imagine moving at high speed towards a cake that you want to eat - fine if you stop when you hit the cake, but difficult if you go straight through it). I suppose it doesn't matter too much though. Most bots stop before they hit vegs anyway
-
Maybe that's a good diet strategy. "You're only allowed to eat as much as you can catch when I throw the cake in your face."
-
Maybe that's a good diet strategy. "You're only allowed to eat as much as you can catch when I throw the cake in your face."
-
Another bug Someone has already sort of noticed it, but I don't think anyone fully realized what this peice of coding could be used for.
'this line below can potentially make waste just disappear! Fix it!
.Waste = .Waste - .mem(824) / 100
For Poison and Venom storing a negative value, while having zero can create waste, for zero nrg expenditure. A simple sharewaste with a plant can then guarantee an infinite food supply.
There also don't appear to be any caps on the amounts being pushed in... Therefore there's no limits on the amounts of waste it can convert to...
On a somewhat sad note, it appears this was responsible for my little "junk code" bots success. The continual waste addition with minimum energy expenditures made unrealisticly ideal conditions for them to evolve under.
On a related note waste does seem to be affecting off bots quicker than normal. In 2.42.2 my bots lasted almost no time at all.
-
On a related note waste does seem to be affecting off bots quicker than normal. In 2.42.2 my bots lasted almost no time at all.
I've noticed this too. A temporary solution is to drastically increase the waste threshold.
-
DB has decided it doesn't want to load my saved files anymore. Maybe someone can check it out. Seems all my saves are getting corrupted, both manually saves and autosaves. When trying to load a save I get this error message: Error while loading sim: invalid procedure call or argument. If I then quit the program I get this error message: unable to save settings: some error occured.
Here's one of my saved sims wich I can't load:
-
It looks like somehow the LastMutDetails of one of the bots got recorded as negative length, which is confusing the program.
The good news is that your robots still exist in the save. The bad news is that the corrupted save is causing all sorts of problems that Eric's going to need to spend some time addressing
-
Lovely. I'll take a look as soon as I can.
-
Here's one of my saved sims wich I can't load:
Attached is the sim loaded and saved with the fixed code (coming in 2.42.4). All but the most recent 2^15 bytes of the mutation details have been stripped and soem bots may have random values in their mutation settings.
It was the same overflow problem with mutation details as described in this thread (http://www.darwinbots.com/Forum/index.php?showtopic=1304&hl=).