- With the entire automotive industry rushing to SAE Level 5 autonomy as specified by the SAE J3016, I have worked with python and MATLAB and simulink for developing algorithms particularly useful for self-driving applications.
- Working on Lane detection, masking regions of interest, converting into grayscale and perfoming hough transformations. Ground truth labeling, traffic sign identification and vehicle tracking concepts are stressed upon while developing code and building models on MATLAB and Simulink. The following images throws light onto the analysis done:
Interstate test image
Interstate test image converting region of interest to gray scale
Interstate test image after region masking
- Worked on developing a lane detection pipeline to detect lane lines with color, gradient and combination threshold techniques.
- Started off by calibrating the camera to achieve camera distortion coefficients and camera matrix to compensate for any tangential and radial distortion in camera images.
- This was further coupled with perspective transform, vehicle offset from lane center, radius of curvature determination.