Re: Re: Braitenberg Vehicle Tutorial - "set-bias" related question
jon klein <[email protected]>
| Newsgroups | gmane.comp.breve |
|---|---|
| Message-ID | <[email protected]> |
On Jul 27, 2004, at 4:46 AM, b n wrote: > The tutorial states that the set-bias function can be > used to "slow down" the vehicle speed by setting a > negitive bias on the Sensors. > > Well, I am able to see this behaviour. But what I > noticed is the following : > > When I use a negative bias of say -1.0, I notice that > the Vehicle NO LONGER "gobbles up" the light source, > as it does when I use a positive bias to for the > sensors. > > Is this expected behaviour or am I missing something. This is the expected behavior -- the tutorial fails to mention that a negative bias will not only slow down or stop the vehicle down, it will also cause the wheels to go backward when the input is strong enough. If you still want the vehicle to move forward, you may want to implement your own custom activation method, which determines how the sensor level gets turned into a wheel velocity. To compute a wheel velocity, the activations from all connected sensors are summed together. Activation is computed like this: activation = activationFunction(sensorLevel) * bias To make sure that wheel velocity never goes negative, you may want to use a normal activationFunction, but give it an upper bound so that the whole equation never gets "too" negative. You would use that in conjunction with a positive natural velocity to get the behavior you describe. Instructions on implementing another activation method are found in the BraitenbergSensor class documentation. -- Incidentally, I've just uploaded an updated version of the tutorial. It doesn't really address the issue you raise, but it does include a number of other improvements. It's available here: http://www.spiderland.org/breve/BraitenbergTutorial.pdf - jon klein _______________________________________________ breve mailing list [email protected] http://www.spiderland.org/mailman/listinfo/breve