Skip to 0 minutes and 2 secondsOK, it's time for an exercise. And this is about programming a robot that works a bit like a Braitenberg vehicle. So here is the web page, and I'll scroll down a bit so we can see the controls. Basically, we've got a Braitenberg vehicle as shown down the bottom, which has got two sensors, left sensor and the right sensor. And they need to have a one if they detect a light, a zero if not. It's then got two sets of three neurons, one which determines the left speed, one determining the right speed.
Skip to 0 minutes and 36 secondsAnd the formulae that are used is that the left speed, as it says there, is the left sensor times the LL neuron, which is that one there, plus the right sensor times the LR neuron, plus the LM neuron. So if the robot, for instance, can't see anything, then the LM neuron will determine how fast it goes. And once you've typed in the different neuron values here, you can press start and see the vehicle move. And as ever, you have the option to have reverse left motor. So once you type something in here, you can see the robot move around. And you can put it back to its initial position with a button there.
Skip to 1 minute and 25 secondsAnd you can specify narrow beams, or wide beams.
Skip to 1 minute and 31 secondsThe two red dots represent the lights, and the sensors can detect these. So first, what you need to do is to find the neurons, so that the robot moves forward when it sees nothing, but slows down and turns towards the light it sees, rather like a passive seeker. Try that with narrow and wide beams. Then define the neuron, so that the robot moves forward when it sees nothing, but speeds up and turns to the light. So it's more like an active seeker. Then define the neurons so that the robot moves forward when it sees nothing, but turns away from the light it sees, which is more like an active phobe.
Skip to 2 minutes and 7 secondsThen if you click the line to follow, which we've seen before, can you now put the neurons in so that the robot follows the line. And it's all done by typing in different values for the neurons. Enjoy.
In the final exercise on the course, you will be able to define the robot speed to specify the value of the neurons in the vehicles. In this video, Richard introduces the simulation and what you need to do to define the neurons so that the robot moves forward or moves away from a light.
© University of Reading