Image Source: http://artint.info/html/ArtInt_51.html
In artificial intelligence, an agent is an entity which perceiving its environment through sensors and acting upon environment through actuators. You can consider agent as independent unit who serve as thinking mind. Thus, the cycle consist of perceive, think and act.
So, what is agent?
The agent is a combination of architecture and program, and it is exist in our daily application. For instance, cell phone, vacuum cleaner, robot, and slef-driving car.
A well-behave agent will maximize its performance measure with percept sequence and build-in knowledge. And it is define in PEAS, Performance, Environment, Actuator, Sensors.
Using self-driving car at street, we will consider factor as below:
Performance: Safety, time to destination, legal drive, comfort
Environment: Roads, other cars, pedestrians, road signs
Actuator: Steering, accelerator, brake, signal, horn
Sensors: Camera, sonar, GPS, speedometer, odometer, engine sensor
At home, using room vacuum cleaner,, we will consider factor as below:
Performance: Cleaness, efficency, distance travel to clean, battery life
Environment: Room, table, wood floor, different obstacies
Actuator: Wheels, different brushes, vacuum extractor
Sensors: Camera, dirt detection sensor, cliff sensor, bump sensor, infrafed wall sensor
Now, let us take a look of one of the agent, Search Agent
A search agent is a goal-based agent which formalized to work towards a goal, and few elements contained in the program which are
– Initial State (Initial point)
– State (Sequence of action)
– Action (eg: left, right, top, bottom)
– Path Cost (cost of each path taken)
The application of the search agent is widely deployed in area below
1) Routing problem
2) VLSI layout
3) Robot navigation
4) Automatic assembly sequencing (eg: Car manufacturing)
In general, search space divided into three region:
– Explored
– Frontier
– Unexplored
Two kinds of search:
Uninformed (The info is not provided)
- (Breadth first search), searching by level, BFS biggest gandicaps = memory requirement and exponential time
- (Deep first search), search into deepest path first,
Informed (The info is provided)
- Uniform Cost Search – cost
Reference source: https://courses.edx.org/courses/course-v1:ColumbiaX+CSMM.101x+2T2017/course/
