One of the Learning Objectives in Unit 3 asks us to "Discuss the creation of robots that employ effectors to interact with their environment", and one of those in Unit 4 asks us to "Describe how sensors can be used to enable robots to perform their functions effectively" and to "Discuss the creation of robots that employ sensors to perform useful functions."

An effector, according to Matarić, is "any device on a robot that has an effect (impact, influence) on the environment." All robots interact with their environment by definition. How do we talk about everything that affects the environment? What is the environment? What doesn't affect the environment?

Sensors provide physical data about the state of the robot’s own body and the immediate environment. Touching the signal-to-symbol problem mentioned in Matarić, this information is converted to a form with which the robot can make intelligent decisions.

As this topic is so broad, I propose that we approach it in the broadest possible way: with physics. To have an effect there must be an energy transfer. The fundamental interactions (forces) are the gravitational and electromagnetic interactions, which produce significant long-range forces whose effects can be seen directly in everyday life, and the strong and weak interactions, which produce forces at subatomic distances and govern nuclear interactions.

Gravitational energy is what we learn in school to call "potential energy", characterized by a ball on top of a hill, ready to roll down. Robots can use potential energy like this with uncontrolled degrees of freedom. That is, once this energy is used to change the robot or environment's state, it can't go back without some other energy input. Something will have to pick up the ball and return it to the top of the hill. The best example I've seen of this is the MorphHex MKII (video link). There are gravitational sensors -- inertial sensors, really -- that can be used, for example, to keep a platform level or a robot upright.

Electromagnetic energy is used extensively in robots, especially electricity. We talk enough about electric actuators, so I won't expand on that, but magnetic and the combined electromagnetic warrants a bit of thought. Infrared, ultra-violet, radio-spectrum, microwave, visible light, x-ray, and so on -- all of these involve photons, packets of electromagnetic energy, and are used for both sensing and affecting the environment. Magnetic effectors are used to move ferromagnetic material for a variety of reasons, such as to sort junk metal from other garbage or to actuate a relay. The earth's magnetic field can be used for rough directional sensing.

The strong and weak forces are not used directly in robotics, unless you count nuclear reactors or autonomous nuclear warheads as robots. Technically, the strong force holds everything together (along with the other forces to a much lesser extent), so it's being used in every interaction involving movement or normal forces (pushing, pulling, etc.). This includes all types of 'bump' switches, whisker sensors, etc.

How do you get something done with these forces? There are few blanket restrictions on what you can do. It all comes down to practicality. What's the easiest, cheapest, fastest, most efficient, most stable, quietest, least straining, ... "best" way to accomplish your goals? Answering a question like that seems to be as simple as pondering the meaning of life. Let's leave these difficulties for more specific situations...

I like the following quote from Matarić:

To design an effective robot, it is necessary to have a good match between the robot’s sensors, task, and environment.