3. (10 points) Suppose your intelligent robot has survived the walls thanks to the fuzzy logic based obstacle avoidance system you developed in the first midterm exam. Now, we add a camera on top of the robot to follow a lane: Projection of the Vlane on the image Lane to follow LEFT SIDE RIGHT SIDE Robot Pixel 0 Pixel n Image Centre In this case, the camera acquires images of the lane and we will use these images to control the motion of the robot similar to what we did previously. We will use a variable d (in pixels) to indicate the distance of the projection of the lane (which will appear as a line on the image) to the centre of the image. So, we will have a rule base as follows: IF d is very near AND the line is on the RIGHT SIDE THEN perform a LOW rotation to the right. IF dis Near AND the line is on the RIGHT SIDE THEN perform a MEDIUM rotation to the right. The crisp value of d will be fuzzified using the input membership given below: (Distance) Very near Near Far 10 20 30 50 80 90 100 Distance (pixels) Then, the amount of rotation is found using the rules in the rulebase using the output membership function below: (Rotation) Low Medium High 20 30 40 60 80 90 Rotation ) At one moment of operation, d value is calculated as 25 pixels and the detected line is on the right side of the image (the membership for this antecedent can be taken as 1.0). Compute the amount of rotation using Mamdani inference method with clipping. (The interval length can be taken as 10). 3. (10 points) Suppose your intelligent robot has survived the walls thanks to the fuzzy logic based obstacle avoidance system you developed in the first midterm exam. Now, we add a camera on top of the robot to follow a lane: Projection of the Vlane on the image Lane to follow LEFT SIDE RIGHT SIDE Robot Pixel 0 Pixel n Image Centre In this case, the camera acquires images of the lane and we will use these images to control the motion of the robot similar to what we did previously. We will use a variable d (in pixels) to indicate the distance of the projection of the lane (which will appear as a line on the image) to the centre of the image. So, we will have a rule base as follows: IF d is very near AND the line is on the RIGHT SIDE THEN perform a LOW rotation to the right. IF dis Near AND the line is on the RIGHT SIDE THEN perform a MEDIUM rotation to the right. The crisp value of d will be fuzzified using the input membership given below: (Distance) Very near Near Far 10 20 30 50 80 90 100 Distance (pixels) Then, the amount of rotation is found using the rules in the rulebase using the output membership function below: (Rotation) Low Medium High 20 30 40 60 80 90 Rotation ) At one moment of operation, d value is calculated as 25 pixels and the detected line is on the right side of the image (the membership for this antecedent can be taken as 1.0). Compute the amount of rotation using Mamdani inference method with clipping. (The interval length can be taken as 10)
