Sandeep Reddy is currently a Robotics AI Intern at Bosch Center for AI, formulating optimal control strategies under sensor uncertainties for Lunar Rover. His research at the University of Washington and beyond is dedicated to robot motion planning and controls, policy learning, and planning under uncertainty, aimed towards robot autonomy. During his tenure at UW Robot Learning Lab as a Research Assistant, Sandeep was part of prestigious DARPA RACER program on building high-speed off-road level 4 autonomous vehicles. Since the summer of 2022, he has passionately contributed to the domain of local planning and controls.
Sandeep’s journey isn’t solely defined by academia; he brings with him a diverse background. His undergraduate years at NIT Warangal, India, culminated in his leadership of a multifaceted SAE BAJA off-road racing team, comprising 25 individuals spanning software, design, suspension, steering, and brakes domains. Before joining UW, Sandeep lent his expertise to Bajaj Auto Ltd., an eminent automobile company in India. Over a period of seven months, he immersed himself in algorithm-based design optimization, fostering innovation. With the skills and experience he obtained so far, he would like to apply them to deal critical problems in projects through creativity.
You are invited to explore his work below.
Projects
You can contact me at sbaddam@uw.edu or LinkedIn
Work Experience
Robotics AI Intern | NASA PFP - Bosch Research, Pittsburgh, PA (Mar. 2023 - Present)
- Performed experiments on Wi-Fi based sensor modality for lunar CubeRover state estimation in smart docking with the lander
- Developed optimal control strategies under sensor uncertainties for effective rover positioning and achieved over 98% success rate
- Incorporated external uncertainty learning with expert demos making EKF quickly adapt to extraterrestrial environments
Research Assistant | Robot Learning Lab, Paul G. Allen School of CSE, UW (Sep. 2022 - Mar. 2023)
- Carrying out Inverse Reinforcement Learning to autotune the planner cost function parameters that can save 3-5 manual hours
- Implemented parallel version for cross-track error in MPC-based local planning that is 115X faster than serial computation
- Developed motion planning debugging tools to understand the planner decisions quantitatively using expert demonstrations
Software and Field Test Engineering Intern | DARPA RACER Program, University of Washington (Jun. 2022 – Sep. 2022)
- Worked on motion planning and control for Level 4 autonomous high-speed off-road vehicle
- Contributing algorithms to safely handle vehicle attitude (roll, pitch) in a model predictive local planner and autonomy status
- Executed a logic to speedup manual takeover by approximately 0.5 sec
Graduate Student Researcher | Ultra Precision Controls Laboratory, University of Washington (Dec. 2021 – Jun. 2022)
- Wrote an obstacle avoidance algorithm that uses 2D LiDAR point cloud to maintain multi-robot formation during navigation
- Worked on A* search and CUDA-based Depth Estimation using Stereovision that is 57X faster than the serial version
- Implemented Feedback Linearization based trajectory planning for non-linear models using self-made control tuning UI
Research & Development, Graduate Trainee Engineer | Bajaj Auto Ltd., Pune, India (Jan. 2021 – Jul. 2021)
- Designed and validated a technique to find the precise cabin volume of any closed car with an accuracy of 98%
- Wrote an optimization algorithm to find best HVAC design parameters and experimented in real-time on different cars which resulted in validation accuracy > 95%
Captain, Designer, Driver– SAE BAJA off-road racing team | NIT Warangal (Apr. 2017 - Feb. 2020)
- Managed a 25-member cross-functional (software, design, suspension, steering, brakes) team in building a light off-road vehicle
- Used IoT and Matlab’s ThingSpeak to develop graphical visualization in getting the running status of the vehicle during race
- Improved chassis design using grid independent technique and achieved overall weight under 150kg maintaining min. fos of 1.8