2.9

## University of Reading

Baxter picking up balls

# Sensing and Computing

As humans, we primarily sense our world visually, using our eyes. Our two eyes help for instance to work out how far away an object is and so it seems sensible to add cameras to robots.

However, associated with our eyes we have advanced brains - these allow us to:

• Interpret what we see
• Distinguish different objects from the scene
• Identify each of the objects
• Cope when one object is in front and hence obscures another

If we want a robot to use cameras we need:

• A computer running complicated programs to process the image
• Lots of memory to store the camera images

So for simple robots a simple ultrasonic sensor may be better.

Even then an actuator is needed to emit the pulse and a sensor is needed to detect the reflected signal. Some processing is needed to deduce the time taken before the reflected signal is received and hence calculate the distance of an object.

Distance = speed of signal * time taken / 2


We divide by two as the signal goes from the robot to the object and then back to the robot, that is the signal travels twice the distance.

But speed of signal in air can change with humidity and temperature - the signal emitted may vary. When signals reach the surface of an object, some surfaces reflect stronger signals than others. Like a mirror reflects light better than other surfaces.

The returned signal will vary depending on the reflective surface so even simple sensors need some complex electronics/computing.