Pranav Shankar Srinivasan
Salt Lake City
Overview
5
5
years of professional experience
Work History
Guidance Navigation and Controls Engineer
Autonomous Solutions Inc
04.2025 - 01.2026
- Implemented C++ regression tests to validate velocity-controller performance and improved baseline tracking metrics, including the mean and standard deviation of normal error, as well as the mean and maximum heading error.
- Enhanced the Extended Kalman Filter (EKF) by incorporating lateral (Y-axis) velocity, improving state-estimation accuracy by 20% in simulation.
- Analyzed and characterized the controls stack's limitations for integration with electric vehicles, including regenerative braking.
- Analyzed Vector Path Control (VPC) behavior and improved trajectory-tracking accuracy by reducing mean and maximum normal error by 35 percent.
- Optimized clothoid parameters by tuning curvature-rate gains, improving kinematically feasible path generation to 90% under tight maneuver scenarios.
Research Engineer
Novus HiTech
04.2023 - 06.2024
- Analyzed traversability mapping (TM) for an autonomous tracked vehicle operating in unstructured terrain (DRDO collaboration). Tuned voxel size and elevation thresholds, achieving 95% traversability classification accuracy.
- Simulated Hybrid A* Lattice planner to find a kinematically feasible path for an ATV with the least time (100 ms) taken to plan. Smoothened the path retrieved from the path planner using nav2_constrained_smoother and developed test cases for the ATV to navigate with different maneuvers.
- Evaluated MPPI with a Regulated Pure Pursuit controller, identifying a peak lateral track error of 1.5 m (0.5 m mean) in high-curvature segments, and reducing the peak error to 0.4 m through controller tuning.
Autonomy Engineer
Ati Motors
06.2022 - 12.2022
- Developed a brake control algorithm for an autonomous cargo vehicle, and performed tests to minimize slip on flat terrain and a 3-degree ramp. Analyzed the braking distance of the vehicle when the vehicle is coupled to a load of 350 kg, and reduced the braking distance by 50 percent.
- Simulated an EKF by fusing IMU (gyroscope) and wheel encoder to detect faulty encoder data. Performed tests on a vehicle to check for vehicle recovery after detecting faulty encoder data using Python.
- Performed tests to minimize oscillations in cross-track error (CTE) with a maximum CTE of 0.20 meters by utilizing a non-linear Model Predictive Controller (MPC) for a velocity of around 1.0 m/s.
Research Associate Intern
Public Invention
05.2021 - 02.2022
- Utilized a Proportional Integral Derivative (PID) control module to control the valves in an oxygen concentrator, which delivers medical-grade oxygen (>90% FiO2) at target flow rates of 5 LPM, 20 LPM and 50 LPM etc.
- Performed mock simulations of the PID controller using C++, at desired pressures between 1 psi and 20 psi, to reduce steady-state error. Optimized the PID controller to make it adaptive to different pressure trajectories and mock pressures.
Education
Master of Science - Mechanical Engineering
Columbia University
New York02.2021
Bachelor of Engineering (B.E.) - Mechanical Engineering
SSN College of Engineering
India06.2019
Skills
- C 11/14
- Python
- ROS 1/2
- MoveIt 2
- Navigation 2
- Point Cloud Library (PCL)
- OMPL (Open Motion Planning Library)
- PyBullet
- SolidWorks
- MATLAB
- SQL
- Eigen
Visa Sponsorship
No visa sponsorship required
Projects
- Incorporated Deep Q Network (DQN) for balancing a cartpole and compared with Double Deep Q Network (DDQN). Achieved 30 percent increase in duration of balance for DDQN when compared to DQN. Analyzed the efficiency of Deep Deterministic Policy Gradient (DDPG) and Twin Delayed Deep Deterministic Policy Gradient (TD3) for autonomous navigation of mobile robots in indoor environments. Observed a decrease in training loss by 30 percent for TD3 compared to DDPG.
- Utilized the MoveIt Task Constructor (MTC) for grasping objects (cylinders and boxes) using a 7 DoF panda arm and analyzed performance characteristics like optimal time taken and optimal cost for allocation of different tasks., Established an increase in percentage of successful plans by 15 percent for motion planning using STOMP (Stochastic Trajectory Optimization for Motion Planning) when compared to CHOMP (Covariant Hamiltonian Optimization for Motion Planning).
- Simulated the motion of a 4 bar linkage. Performed inverse dynamics control of a kinova jaco arm., Trained a 5DoF Panda arm to open doors using TD3 and achieved a training accuracy of 93 percent over 100 episodes.
- Designed and fabricated a biped prototype having a height of 1.5 feet and actuated by 6 Lewansoul LX-16A motors., Tuned the gait for the legs to achieve a walking speed of around 1.0 cm/second.
- Designed a 2 RPS - 1 PRS parallel actuated structure using Solidworks and tested the device to do flexion/extension movement and lateral bending movement of the neck using MATLAB. LABVIEW was utilized to input signals to the actuators., Implemented zero force control to validate forward kinematics (FK) by comparing the FK solution of the parallel actuated mechanism with the motion of the wearable device.
Publications
SSN College of Engineering,
A 4 DoF robot for sorting objects, 06/01/19, 10.1063/5.0004429.
Utilized an image processing algorithm which works on averaging the hue and saturation values in the pixels to sort colors., Designed and fabricated the robot after analyzing the stress induced on the robot while doing the grasping task.
Timeline
Guidance Navigation and Controls Engineer
Autonomous Solutions Inc
04.2025 - 01.2026
Research Engineer
Novus HiTech
04.2023 - 06.2024
Autonomy Engineer
Ati Motors
06.2022 - 12.2022
Research Associate Intern
Public Invention
05.2021 - 02.2022
Master of Science - Mechanical Engineering
Columbia University
Bachelor of Engineering (B.E.) - Mechanical Engineering
SSN College of Engineering