Me

About Me

Hi, I am Siddharth

I am a MSCS Graduate from UC San Diego. My reseach involves achieving safe and optimal planning in the multitude of platforms I work with. My work deals with building autonomous infrastructure and algorithms across all levels of the stack. I have worked with planners, perception, localization and simulation. I lead, design, build, test, integrate and maintain codebases for robust and safety critical autonomous vehicles

Resume

teamwork
IAC Match Overtaking

Work Experience

TUM vs ART
Lead Software Engineer and Stack Developer

Click on the image to learn more

Lead Software Engineer and Stack Developerclose
  1. On-site lead responsible for deployment and integration of all software on our team's Indy Autonomous Challenge racecar
  2. Developed C++ ROS 2 driver to connect to N instances of Luminar Iris, control parameters and retrieve sensor health and pointcloud information
  3. Developed generator for C++ ROS 2 CAN drivers given DBC files. Used to generate drivers for Raptor, Motec, Spacedrive, TTPMS and Brake Temps in Indy Autonomous Challenge
  4. Optimized drivers to reach 30-60% reduction in CPU load and 70%+ reduction in latency
  5. Developed CI testing pipeline using Github Actions
  6. Developed C++ ROS 2 Path Planner that takes behavior and an occupancy grid to generate trajectories to follow. Employs Kinodynamic Curvilinear Sampling and Parallel Path Generation for efficiency
  7. Developed C++ ROS 2 Behavior Planner that takes into account current environment conditions, perception, race control commands and general goal of the race to make decisions and execute. Modified for race rules every season. Based off Behavior Trees
  8. Developed interface to open source Unity based OSSDC simulator. Made use of Python API to create custom scenarios and multi ego environments
  9. Developed co simulation option using ChassisSim by integrating into MATLAB and Simulink to load into Unreal Engine based Carla
  10. Developed Scenario Runners for simulations using Bazel and gRPC microservices to interface with varying autonomy and simulation services
  11. Developed C++ ROS 2 Kalman Filter for localization that fuses wheel speeds, IMU and GPS to achieve highly reliable estimations during GPS dropouts via dead reckoning
  12. Developed C++ baseline lidar preprocessing and detection stack using Autoware.Auto
Go Kart
Technical Lead and Software Engineer

Click on the image to learn more

Technical Lead and Software Engineerclose
  1. Competed in evGrandPrix, ICRA, F1TENTH and DIY Robocars competitions
  2. Lead for the evGrandPrix Autonomous Challenge 2022
  3. Integrated Indy Autonomous Challenge Solution to go kart
  4. Developed baseline stack that makes use of ROS2 Nav2 to navigate in racing environments. Uses costmaps and GPS to navigate and stay within bounds while avoiding obstacles
HDSI
Teaching Assistant

Click on the image to learn more

Teaching Assistantclose
  1. Tutored for Introduction to Autonomous Vehicles, Principles Of Data Science, Programming and Basic Data Structures for Data Science, Data Structures and Algorithms for Data Science, Practice and Applications of Data Science, Machine Learning: Representation and Data Science Capstone - Robotics
  2. Ran discussion sections in person for entire section of class
  3. Answered student questions and provided office hours for interactive feedback
  4. Helped developing assessments in both direct examination and homework
Go Kart Run

Projects & Demos

1/10th car
Camera Based Mapping and Navigation

Click on the image to learn more

Camera Based Mapping and Navigationclose

A ROS based stack integrating RTABMAP, Detectron2 and the F1Tenth platform with some tuning done on the Intel Realsense D455. Featured on Nvidia Jetson Community Projects as Project of the Month March 2022. Read more here

Smartboard
Smartboard

Click on the image to learn more

Smartboardclose

Enabling verbal communication with just the use of head movements. Read more here

F1Tenth Sim
IROS 2020 F1Tenth 3rd Place Winner

Click on the image to learn more

IROS 2020 F1Tenth 3rd Place Winnerclose

Won 3rd place in IROS 2020 F1Tenth Race as a one man team. Solution integrating 2D Lidar to achieve robust navigation through a race track