Performance Measurement in Vacuum Cleaner World

Exercise 2.11:

Objective:

Implement a modular performance-measuring environment simulator for the vacuum-cleaner world to facilitate easy modification of sensors, actuators, and environment characteristics.

Methodology:

1. Simulation Environment Setup:

   • Design a modular simulator for the vacuum-cleaner world.
   • Define parameters such as environment size, shape, dirt placement, etc.

2. Agent Implementation:

   • Implement an agent with sensors and actuators for navigating and cleaning the environment.

3. Performance Measurement:

   • Measure performance metrics such as time taken to clean the environment, number of actions performed, etc.

4. Modularity Testing:

   • Test the simulator's modularity by easily changing environment characteristics and agent behaviors.

Exercise 2.14:

Objective:

Investigate the performance of different types of agents in a modified vacuum environment with unknown geography and initial dirt configuration.

Methodology:

1. Simple Reflex Agent Evaluation:

   • Analyze if a simple reflex agent can be perfectly rational for the modified environment.

2. Randomized Agent Function:

   • Design a simple reflex agent with a randomized agent function and measure its performance compared to a deterministic simple reflex agent.

3. Performance Testing:

   • Evaluate the performance of both agents on various environments.

4. Environment Design:

   • Design environments where the randomized agent performs poorly to showcase its limitations.

5. Reflex Agent with State:

   • Implement a reflex agent with state and measure its performance against a simple reflex agent.

6. Rational Agent Design:

   • Explore the possibility of designing a rational agent based on the reflex agent with state.

Video presentation:

https://www.youtube.com/watch?v=NWyIN0GGA58&ab_channel=wushi