Hi... This is Abhay Karade
About
Hello, and welcome to my profile! I am a passionate creator with strong technical skills and a demonstrated history of technical project management. I have a keen interest in working on challenging problems in the field of perception and vision-based robotic manipulation that can add value to businesses.
My primary area of focus is in developing innovative solutions to complex problems. I possess a deep understanding of various technical fields and enjoy leveraging my knowledge to create new solutions.
Aside from my technical expertise, I am known for my exceptional people skills. I believe that effective communication and collaboration are essential to achieving success in any project.
In my free time, I enjoy reading books, and Simon Sinek is one of my favorite authors. His book "The Infinite Game" has greatly inspired me and has helped shape my approach to problem-solving.
Projects
Visual Adaptive Control of 7 DoF Robot Using Keypoints
Developed and tested deep learning models to detect key points on 7 DoF robots.
Implemented a software pipeline to get camera extrinsic from detected key points without using fiducial Markers
Camera Calibration through Keypoint Detection on 7 DoF Robot
Developed and tested deep learning models to detect key points on 7 DoF robots.
Implemented a software pipeline to get camera extrinsic from detected key points without using fiducial Markers
OffroadNet
OffroadNet is a collection of model architectures that focus on the task of improving performance of traversable path detection in unpaved roads. We present the results of our experiments on the Robot Unstructured Ground Driving (RUGD) Dataset and train models to specifically focus on path detection performance. We use mIoU and mAcc as our metrics to compare the performance
of different models. OffroadNet (based on PSP- Net) and ResNet101 is one of the best combi- nation for the offroad path detection because of
it’s PPM module and fine tuning based on the Cityscape dataset.
High-Speed defect detection ‘Uncut Cap’
Developed automatic defect detection and rejection mechanism at a machine speed of 2500 capsules/min.
Used inputs from semiconductor LASER and absolute encoder for positional reference of the defect.
Total of $ 800,000 of savings for four Indian plants.
Lip movement detection under surgical mask
To identify the speaker on any video call with a mask, I created a Dataset of Surgical Mask and used Mask R-CNN for segmentation. Used sparse optical flow to get flow vectors of features of interest. Detected speaker with a webcam, speaking at a speed of 120 WPM
Gallary
