SRI MANIKA MAKAM

****, **** ******, *** ***, College Park, Maryland-20740 301-***-**** adjv1x@r.postjobfree.com Relevant URLs: www.linkedin.com/in/manikamakam, https://github.com/manikamakam EDUCATION

Professional Master of Engineering Robotics GPA: 3.811/4.0 Expected May 2021 University of Maryland College Park, MD, USA

Coursework: Planning for Robots, Decision Making for Robots, Perception for Robots, Statistical Pattern Recognition, Computer Processing of Pictorial Information, Robot Modeling, Control of Robotic Systems, Software Development for Robotics Bachelor of Engineering Electronics and Communication GPA: 8.37/10 May 2019 Chaitanya Bharathi Institute of Technology Hyderabad, India SKILLS

Programming Languages: C, C++, Java, Python, VHDL, Verilog, Embedded C. Others: ROS, Arduino, Raspberry Pi, Linux, MATLAB, SolidWorks, Unity, TensorFlow, PyTorch, Keras, scikit-learn, Numpy, Pandas, OpenCV, CMake, Git, BitBucket, JIRA, Confluence, Travis CI, Coveralls (Test Coverage), Valgrind, Agile Development, CUDA. EXPERIENCE

Computer Vision Engineer June 2020-Dec 2020

SimInsights, Inc Intern, Remote

• Generated synthetic dataset from CAD model in Unity; Trained YoloV3, DetectNet and FasterRCNN with with Isaac SDK; Performed domain randomization on synthetic dataset to improve accuracy of trained networks on real-world dataset.

• Achieved sym2real transfer learning in pose estimation by training PoseCNN.

• Developed an AR video analytics system with NVIDIA Deepstream on Jetson Nano.

• Worked on a VR planning system with PDDL (Planning Domain Definition Language) formulation. Research Student May 2020-Present

Under the guidance of Dr. Derek A. Paley University of Maryland, College Park

• Developed a system with which a drone can be controlled by gesture (EMG patch electrodes on hand) and is equipped with haptic feedback through vibration motors on hand.

• Working on extending this system to a swarm of crazyflie drones using LPS, flow deck and multi-ranger deck. Graduate Student Member October 2019-March 2020

Autonomous Micro Air Vehicle (AMAV) team of UMD University of Maryland, College Park

• Developed a pay load release mechanism for the drone by 3D printing the parts.

• Worked as a member of the Autonomy team and developed the path planning module for an Autonomous UAV.

• Secured second place in Vertical Flight Society Micro Air Vehicle Student Challenge in May 2020. PROJECTS

Structure from Motion

• To reconstruct a 3D scene and simultaneously obtained the camera poses of a monocular camera w.r.t the scene.

• Implemented this using the traditional geometric vision approaches and deep neural networks. Creating one seamless panorama image from sequence of images

• Developed the pipeline of corner detection, ANMS, feature matching, RANSAC, homography estimation and warping.

• For same task, implemented a supervised learning model (HomographyNet) and an unsupervised network with TensorDLT. Face Swap

• Implemented using two traditional methods – Face warping with triangulation and Thin Plate Spline.

• Achieved the same with supervised encoder-decoder neural network which outputs a full 3D mesh of the face. Simulation of a fully functional self-driving car using applied deep learning

• Performed Lane Detection using Hough Lines; Achieved autonomous driving using behavioral cloning.

• Implemented a LeNet model to classify 43 road symbols. Implementation of Decision-Making Algorithms in Open AI Highway Environment

• Implemented Model based RL, Deep Q-Learning, Soft Actor-Critic and Deep Deterministic Policy Gradient. Habitat Challenge 2020 – PointNav

• Integrated ORB-SLAM2 with Proximal Policy Optimization (PPO) baseline network and both are trained simultaneously to deal with the absence of GPU+Compass sensors; Developed an efficient point navigation system. Simulation of ROS based navigation system for a custom robot

• Added laser and IMU sensors to custom robot; created map using SLAM; developed RRT* planning algorithm and implemented lateral and longitudinal control of the robot to achieve car-like locomotion. Design of Feedforward Model Predictive Controller for Path Following of an AGV

• Compared MPC and MPC with feedforward for path following of an autonomous vehicle in MATLAB. First place in “Technica 2019” challenge by ASRC Federal

• Task was to identify features on a Mars like surface through a drone feed. Created a data set from scratch, trained neural network using TensorFlow and detected features using OpenCV.

Contact this candidate