Junho Ham

Background

About Junho Ham

I am developing a foundational AI model for autonomous driving.@MAGNA 2023 Built the panoptic segmentation training and inference pipeline Built a centi-meter-level 3D reconstruction pipeline from distorted fisheye images with OpenCV and open model for dense optical flow Developed the transformer model for depth estimation of detections and road-topology prediction @MAGNA Implemented the Longitudinal Controller for ACC, AD from scratch Implemented the Lateral Controller for LKAS, ALC, AD from scratch Implemented the Longitudinal Estimator for speed, ACC, hill angle from scratch Implemented the Lateral Estimator for steering angle, yaw rate, yaw bias from scratch Implemented the ADAS Sensor Fusion from scratch: 1 Front Camera, 5 Radars Developed the explicit path planner, obstacle avoidance, and decision-maker from scratch Developed a multi-camera perception stack with a deep neural network Optimized and deployed the DNN on the edge device (TensorRT, Xavier, Orin) Optimized and deployed the fusion tracker and planner into the edge device Integrated my tracker, planner, deep learning stacks to the prototype delivery bot. https://youtu.be/MjYz1BeGUzI Developing several deep learning models: semantic/panoptic segmentation, and something with transformer.@HYUNDAI MOBIS Developed the ADAS ACC Development, especially control algorithm Developed the Map-Enabled ACC, especially control algorithm Developed the Path tracking autonomous driving prototype from scratch and demonstrated it Developed the Object tracker for AD, Radar/Camera Fusion, and control Proposed the efficient HD Map based localization with Kalman filter schema HD Map based localization and AD with the localization @LG Electronics Developed the object tracker with pixel-level detection and sub-pixel level image scale detection for front automotive camera