Lecture 06 - Evolutionary Robotics

Question 1

Evolutionary robotics

How much a priori knowledge does the Aggregate fitness function have?

Answer 1

Low

Question 2

Evolutionary robotics

How much a priori knowledge does the Behavioral fitness function have?

Answer 2

Medium

Question 3

Evolutionary robotics

Describe the Tailored fitness function.

Answer 3

It’s a combination of the Aggregate and Behavioral fitness functions.

Question 4

Evolutionary robotics

Describe the Functional incremental fitness function.

Answer 4

It’s a sequence (over the evolution) of fitness functions, AFF, then BFF

Question 5

Evolutionary robotics

For soft robotics, what is morphology?

Answer 5

Morphology is what the body looks like, that is, how the voxels are aggregated.

Question 6

Evolutionary robotics

In soft robotics, what is the controller?

Answer 6

The controller is basically “the mind”. It controls how the voxel volumes change over the time.

Question 7

Evolutionary robotics

In soft robots, what is evo and what is devo?

Answer 7

Evolution and development.

Question 8

Evolutionary robotics

What is Transferability?

Answer 8

Whether a solution can be used in another situation, e.g. can simulated robots be used in the real world?

Question 9

Evolutionary robotics

What is the name of this concept?

Whether a solution is applicable elsewhere, e.g. simulated -> real world.

Answer 9

Transferability.

Question 10

Evolutionary robotics

How do you measure transferability?

Answer 10

Sample the real world performance, compute real/simulated distance, and compute the distance. Lower distance = higher transferability.