Skip to content
View infinityengi's full-sized avatar
1 follower · 58 following

Achievements

Achievement: Pull SharkAchievement: Quickdraw

Achievements

Achievement: Pull SharkAchievement: Quickdraw

Highlights

  • Pro

Block or report infinityengi

Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
infinityengi/README.md

Hi — I'm Om Prakash Sahu 👋

Machine Perception • Autonomous Systems • Embedded Automotive

  • ✅ Focus areas: Perception (LiDAR/Camera), 3D detection & segmentation, SLAM/localization, control (MPC), and automotive systems engineering (ISO 26262, V-Model).
  • ✅ Stack: Python, C++, ROS1/2, TensorFlow/PyTorch, PointPillars, OpenCV, Docker, CI/CD, Plotly, Jupyter Lab.

Key skills & technologies

Python C++ ROS LiDAR Deep Learning Docker TensorFlow PyTorch OpenCV MPC

The Notebooks and Docker badges on each project indicate quick reproducibility for live demos.

Projects

Legend: 📒 = Notebook(s) • 🐳 = Docker • ⚙️ = C++/embedded • 🤖 = ROS • 🔬 = Research/experiments • 🏁 = Demo/reproducible

Visual Lane Following Robot (ACDC) — ROS on Jetson Nano 🤖🏁

Short: ROS system for lane detection, localization + steering (MPC) running on NVIDIA Jetson Nano; includes simulation + real-world tests.

Link: https://github.com/infinityengi/visual-lane-following-robot-acdc

Tags: ROS C++ Python Jetson Nano MPC Simulation Robotics.

goal-driven-td3-nav — TD3 + ROS Noetic + Gazebo 🤖🐳🔬🏁

Short: TD3-based deep RL for goal-driven mobile robot navigation in ROS Noetic + Gazebo. Trains policies from Velodyne LiDAR using PyTorch, Docker-ready, TensorBoard logging.

Link: https://github.com/infinityengi/goal-driven-td3-nav

Tags: ROS Gazebo PyTorch TD3 LiDAR Docker TensorBoard Noetic Research.

A. Perception & 3D Sensing

1. semantic-image-segmentation — image segmentation starter kit 📒🐳

Short: Reproducible starter kit: data pipelines, color→class mapping, U-Net baseline, augmentations, training & export (SavedModel/ONNX/TFLite).

Link: https://github.com/infinityengi/semantic-image-segmentation

Tags: U-Net Data Pipeline Augmentation TensorFlow PyTorch Experiment Tracking Export.

2. point-cloud-semantic-segmentation — LiDAR semantic segmentation 📒🐳🔬

Short: TensorFlow-based pipelines for point-cloud segmentation, cross-modal label transfer, and interactive Plotly visualizations.

Link: https://github.com/infinityengi/point-cloud-semantic-segmentation

Notebooks: 1_Semantic_Point_Cloud_Segmentation.ipynb • 2_Boosting_Semantic_Point_Cloud_Segmentation.ipynb

Tags: Point Cloud Semantic Segmentation TensorFlow Augmentation Plotly Docker.

3. 3D-object-detection — LiDAR-based detection & visualization 📒🔬

Short: Reproducible 3D detection pipeline (PointPillars) on KITTI with end-to-end notebooks for dataset prep, training/inference, and visualization.

Link: https://github.com/infinityengi/3D-object-detection

Tags: 3D Object Detection PointPillars KITTI LiDAR Visualization Python Notebooks.

Quick-run: Preprocessing notebook, anchor & hyperparameter inspection, 2D/BEV visualizers.

4. Semantic Grid Mapping — BEV & occupancy grid mapping 📒🔬🐳

Short: Reproducible framework for grid-based environment representation using camera and LiDAR; demonstrates semantic segmentation → BEV and occupancy grid mapping (PointPillars baseline).

Link: https://github.com/infinityengi/Semantic-Grid-Mapping

Highlights: pillarization, evidential prediction head, IPM + multi-camera stitching for 360° BEV.

Tags: LiDAR BEV Semantic Segmentation PointPillars Computer Vision Python TensorFlow Docker Notebooks.

Quick-run: notebooks/01_pointcloud_grid_mapping.ipynb • notebooks/02_camera_grid_mapping.ipynb

5. inverse-perspective-mapping-cpp — IPM in C++ / OpenCV ⚙️

Short: C++ implementation of Inverse Perspective Mapping for BEV generation, with configuration and OpenCV backend. Good for embedded/real-time tasks.

Link: https://github.com/infinityengi/inverse-perspective-mapping-cpp

Tags: C++ OpenCV IPM BEV Real-time Embedded.

Quick-start: conda env or pip + Jupyter notebooks (notebooks/ quick-run cell).

B. Localization & Mapping

6. AutoSeg-Localization — segmentation + localization research framework 📒🐳🔬

Short: Reproducible framework combining semantic segmentation with vehicle localization; Dockerized, curated notebooks, and experiment-tracking ready.

Link: https://github.com/infinityengi/AutoSeg-Localization

Tags: Localization Semantic Segmentation Docker Experiment Tracking Notebooks Research.

C. Control, Systems Engineering & Safety

7. control-perception-hubs — MPC, Robust Control, Sensor Fusion learning hub 📒🔬

Short: Knowledge hub with tutorials, reference implementations and small reproducible projects across MPC, robust & networked control, and sensor fusion.

Link: https://github.com/infinityengi/control-perception-hubs

Tags: MPC Control Theory Sensor Fusion Tutorials Notebooks.

8. v-model-automotive-portfolio — V-Model for automotive projects 📒⚙️🏁

Short: Systems-engineering toolkit that maps artifacts to V-Model phases; includes lane-keep-assist case study with traceability, tests, and firmware examples.

Link: https://github.com/infinityengi/v-model-automotive-portfolio

Tags: V-Model ISO 26262 AUTOSAR HIL/SIL Systems Engineering Traceability.

9. functional-safety-iso26262 — practical ISO 26262 notes & templates 📒

Short: Concise notes, tutorials and worked case studies for ISO 26262 — includes HARA exercise, management, HW/SW guidance and templates for safety cases.

Link: https://github.com/infinityengi/functional-safety-iso26262

Tags: ISO 26262 Functional Safety HARA Safety Case Automotive.

D. ADAS

10. ADAS-HandsOn-Repo — ACC, AEB, LKA mini-projects & CI-friendly examples 📒🏁

Short: Hands-on repositories for common ADAS features with reproducible notebooks and CI examples.

Link: https://github.com/infinityengi/ADAS-HandsOn-Repo

Tags: ACC AEB LKA CI Notebooks ADAS.

11. Lane Detection Using K-Means Clustering — OpenCV pipeline ⚙️

Short: Color clustering + polynomial fitting pipeline for robust yellow lane detection under challenging conditions.

Link: https://github.com/infinityengi/Lane-Detection-Using-K-Means-Clustering

Tags: OpenCV Computer Vision K-Means Lane Detection.

12. Curved Lane Detection (Sliding Window) — histogram + sliding window method ⚙️

Short: Classic sliding-window lane detection with polynomial fitting, tested under varying curvature & lighting conditions.

Link: https://github.com/infinityengi/curved-lane-detection-sliding-window

Tags: Lane Detection Sliding Window Polynomial Fit.

E. Knowledge & Productivity Tools

13. Professional Workflow Notes — workflow templates & docs 📒

Short: Templates and a repeatable workflow from ideation → delivery: docs, diagrams, pseudocode drafts, and AI-assisted engineering docs.

Link: https://github.com/infinityengi/professional-workflownotes

Tags: Workflow Templates Docs Engineering.

📫 Contact

Pinned Loading

  1. goal-driven-td3-nav goal-driven-td3-nav Public

    TD3-based deep RL for goal-driven mobile robot navigation in ROS Noetic + Gazebo. Trains policies from Velodyne LiDAR using PyTorch, Docker-ready, TensorBoard logging.

    Python

  2. point-cloud-semantic-segmentation point-cloud-semantic-segmentation Public

    Semantic segmentation of LiDAR point clouds using TensorFlow and Jupyter notebooks. Includes reproducible Docker environments, baseline models, data preprocessing, augmentation strategies, and lite…

    Jupyter Notebook

  3. semantic-image-segmentation semantic-image-segmentation Public

    Reproducible starter kit for semantic image segmentation (notebooks, Docker, augmentation, common baselines (e.g., U-Net), and experiment tracking).

    Jupyter Notebook

  4. visual-lane-following-robot-acdc visual-lane-following-robot-acdc Public

    Course project at ika, RWTH Aachen. ROS-based lane-following system for a 1:10 autonomous vehicle using Jetson Nano, Gazebo simulation data, and integrated perception, localization, and motion-plan…

    Jupyter Notebook