Nagarjun Vinukonda

Location: OH, USA

Tools: ROS, C++, Eigen Lib, Dijkstra, KD-Tree, Rviz

• Developed Route Planner stack for Autonomous vehicle, its simulations on Rviz and Performing tests on real vehicle
• Implemented Adaptive destination selection for reference planning and develop planning stack diagnosis system
• Tuning Lookahead distance for pure pursuit and Stanley Controller and test on Thordrive vehicle

Location: CA, USA

Tools: Python, Raspberry pi, Pycoral TPU, Linux, Tensorflow

• Midea Group Emerging Technologies is a startup from Midea Group (Fortune Global 500 Company).
• They work on developing mobile robotic lawn mover and I have worked in Perception Team.
• Developing robot simulations, perception system with NN & train in embedded systems for mobile robotics application
• Design, implement navigation algorithms, setup experimental process & develop test scripts for robotics SLAM system

Tools: Python, PyTorch, CUDA

• Implemented DQN and Dueling DQN in PyTorch with experience replay and target networks for navigation with LiDAR observations.
• Designed reward shaping around goal heading, distance, obstacle proximity, and goal/collision outcomes; trained in Gazebo and deployed the learned policy to a physical TurtleBot.

Illustrative TurtleBot-style simulation with moving obstacles; scripted visualization, not a recorded DQN or Gazebo rollout. GIF ↗

Tools: ROS, Gazebo, C++

• Hardware and software implementation of Turtlebot path tracking system using PID controller
• Experimented on single and multi-goal points determining steering control at each instant of time while navigation

Original project demonstration. Source ↗

Tools: ROS, Python, OpenCV, Gazebo

• Implemented Autonomous, Semi-Autonomous functions for moving 7 DOF Baxter robot arms in ROS and designed gazebo environment
• Created ROS Nodes to build Meshed, Cone and Take Control Methods that works the Semi-autonomous function reducing human fatigue while teleoperation
• Programmed Aruco Marker detector for identifying cups in gazebo environment using OpenCV bridge

Tools: ROS, Gazebo, PCL, C++, Python, Solid Works

• Implemented simulation environment of Lidar laser scan and published its data in Point cloud using ROS and Gazebo
• Investigated DMPs and assisted to train the DMPs generating trajectories for determined joint angles
• Implemented path planning approach algorithm for stair climbing and generating step trajectories
• Operated Motion capture lab to collect the gait data of 15 subjects using Vicon Nexus for generating joint angles
• Worked on LiDAR data acquisition, line segmentation of point cloud data using RANSAC and visualization of staircase using ROS

Tools: ROS, Gazebo, C++

• Implemented the state-of-the-art RGB-D SLAM algorithm for 3-D map building of indoor spaces
• Generated 2D occupancy grids and 3D point cloud data while navigating mobile robot in gazebo using RTAB mapping

Tools: OSM, C++

• Constructed a Route Planner based on A* algorithm on the OpenStreetMap framework.
• Developed the project using the IO2D library with C++14 on linux based system

Tools: Python, Tensorflow, GCP console

• Performed 3D object detection on the KITTI dataset with the goal of reduced computation time and accuracy
• Experimented with using SqueezeDet in Frustum PointNet architecture for 2D detection instead of fine-tuned Fast R-CNN

Tools: OpenCV, Python

• Extracted image features using SIFT & SURF, Hough transforms, Top-hat filter algorithms compared statistically with YOLOv3
• Integrated Darknet YOLOv3 Deep Neural Network for robust real time object detection and SVM for status recognition
• Built an HSV Classifier for real-time classification of Traffic Lights for Autonomous Vehicles

Tools: C++, Linux

• Implemented a system monitor to track the cpu utilization, ram usage, activity time and base-command.
• Compatible with all Ubuntu Distributions, tested on Ubuntu 16.04 and Ubuntu 18.04
• Based on Object Oriented Programming using C++14

Tools: MATLAB, Simulink, C++

• Worked with Prof. Xiangrui Zeng for creating a behaviour planner and extending on CARLA
• Implementing behaviour prediction engine with MPC controller and Integrating Sensor fusion modules to detect multi agents
• Determining cost functions to create constraints for decision making module in Lane changing task with MIOs
• Evaluating Motion Metrics for collision detection in Simulink using stateflow diagram and triggering decision to change lane.

Tools: ROS, Gazebo, C++, Kalman filter, AMCL, Eigen, OpenCV

• Worked at Human-Inspired Robotics(HIRo) Lab with Prof. Zhi Li on Velocity Obstacles method ORCA (Optimal Reciprocal Collision Avoidance) algorithm
• Implementing state of the art method ORCA for social aware navigation of freight robot in hospital environment
• Modelling & Integrating human and obstacle tracking modules for Multi-Human and robot Collision avoidance detection.
• Integrating RVO2 libraries into ROS environment and tunning & testing ORCA hyperparameters for robust navigation

Tools: MATLAB, C++, Openvibe, LSL, psychtoolbox, Chestnut PCB

• I am advised by Prof.Eduardo Miranda to extend Research on Computer-Music with Robotics
• Developed a BCMI system that enables people with severe motor impairment to play music using prosthesis
• Programmed and modelled an integrated system that maps visual evoked potentials obtained to Brunel hand using MATLAB
• Analysed EEG signals in Openvibe and implemented noise filters using digital signal processing methods
• Researched and implemented Canonical Correlation Analysis algorithm to compute real-time mapping of EEG signals

Original project demonstration. Source ↗

• Educated and exposed on the working mechanism of Section 3 of surface-to-air Akash missile and Brahmos missile, Electro-Optical Director, Stabilized Electro-Optical Sight and wash timers

• Exposed to work mechanisms of pressure sensors used in power window system along with integrated circuits of radars and industrial electronics

• Published at International Journal of Mechanical Engineering & Technology, ISSN Online: 0976 – 6359, Vol.9, Issue 12, December 2018
• The paper presents various methods to improve automobile radiators efficiency and testing methods to validate the performance of radiators

• To study BCI methods and develop a real-time integrated system that connects brain waves to prosthetic arm enabling severe motor impairment patient to play musical instruments using prosthesis

• Competed at Lockheed Martin Aeronautics Co. Lakeland, as Vice-Captain of Team: Aero-VIT -- Regular Class: Aero-design competition
• Design of 144 inches wingspan aircraft, 18lbs payload capacity for carrying 24 passengers (Tennis balls).
• Implemented algorithm for prediction of lift and drag using Simulink along with Structural and fluent analysis of aircraft using Ansys
• Establishing successful test flight along with participation in competition securing second position in Asian-pacific region.

• Competed at Anna University, as Vice-Captain of Team: Flying Inc. -- Micro Class: Aero-design competition
• Design and Analysis of 36 inches wingspan radio-controlled aircraft with successful flight test and participation in competition

• The course focuses on main planning tasks in autonomous driving, including mission planning, behavior planning and local planning.
• Implemented hierarchical motion planner to navigate through a sequence of scenarios in the CARLA simulator: Avoiding static vehicles, following a lead vehicle and safely navigating an intersection.
• Calculated time to collision(static), computed velocity profiles, designed state machine that transitions between lane following, deceleration to the stop sign for safe navigation.
• Generated occupancy grid using lidar scanner measurements, iteratively constructed a probabilistic occupancy grid from log odds updates.

• The course Control of Mobile Robots focuses on the application of modern control theory to the problem of making robots move around in safe and effective ways
• As part of course I have wroked on implementation of project strutctre in Simulink for calculating angular speed of left and right wheel with odometry and wheel encoders, PID regulator.

• Got broad idea of Machine Learning Algorithms, Unsupervised and Supervised learning, bias/variance theory, SVM's and numerous case studies and applications of learning algorithms to build smart robots (perception, control), Computer Vision and other areas.