Exercise 1: The Scale-Invariant Feature Transform (SIFT)

• Download the following files and untar them in the same directory using tar -xzf exercise_1.tgz.
  • exercise_1.tgz containing some files for the exercise
  • sift.tgz containing Andrea Vedaldi's SIFT package.
  • idol_kth.tgz containing image sequences from the KTH-IDOL2 database.
• Read the exercise description: exercise_sift.pdf.
• my_fspecial.m
• my_rgb2gray.m
• Use I = double(I)/256; instead of I = im2double(I);
• Use conv2(I, K, 'same'); instead of imfilter(I, K);

Exercise 2: Premature convergence in particle filter

• Download exercise_2.tgz and untar.
• Read the exercise description.

Exercise 3: Simultaneous Localization and Mapping (EKF)

• Download exercise_3.tgz and untar.
• Read the exercise description.
• SLAM Tutorial: part 1
• Additional exercise:
  Adjust the control noises by increasing the noise on the velocity and gamma steering parameters

  sigmaV= 0.3; % m/s
  sigmaG= (3.0*pi/180); % radians
  

  and decouple Q by setting it independently of sigmaV and sigmaG try values between 0.3 and 1.0 and see the effect.
  Do the same for sigmaR and sigmaB and the R matrix

Exercise 4: Full-Day Assignment

There are four assignments that you can choose from. Descriptions can be found here.