Parth N. Shah

Hey! I’m Parth, welcome to my small space on the internet.

I’ll be joining the MS in Robotics Systems Development (MRSD) at Robotics Institute, Carnegie Mellon University for it’s 2023-2025 cohort.

My interests revolves around making full-stack autonomy solutions that can be deployed in real-world settings. I plan on exploring the domains of Robotic Perception and SLAM in the coming years.

Before joining CMU, I worked as a Research Associate at Dr. K Madhava Krishna’s Robotics Research Center @ IIIT-Hyderabad on developing a Autonomous Driving Stack and Learning-based Planning. Prior to this, I worked as a Robotics Software Engineer at Rightbot, where we developed robots for nano-fulfillment warehouses.

In 2022, I graduated with a Bachelor of Technology degree majoring in Computer Engineering. During my undergrad, I was associated with Orion Racing India - Formula Student Team as a Driverless Engineer working. I was also an Embedded Engineer with Team KJSCE Robocon.

On the fun side, I play table-tennis, go running and binge a lot.

news

Aug 10, 2023	Pittsburgh, here I come!! I am super excited to join the MS in Robotics Systems Development (MRSD) program at Robotics Institute, Carnegie Mellon University.
May 25, 2023	Paper “NeuroSMPC : A Neural Network guided Sampling Based MPC for On-Road Autonomous Driving” accepted at IEEE Case 2023.
Mar 15, 2023	Accepted the offer of joining MS in Robotics Systems Development (MRSD) at RI, CMU!
Aug 17, 2022	Started working at Robotics Research Center, IIIT-Hyderabad.
Sep 15, 2021	Joined Rightbot as a Robotics Software Engineer.

selected publications

2023

IEEE CASE-2023

NeuroSMPC : A Neural Network guided Sampling Based MPC for On-Road Autonomous Driving

Kaustab Pal, Aditya Sharma, Mohd Omama, and 2 more authors

Jun 2023

Abs HTML PDF

In this paper we show an effective means of integrating data driven frameworks to sampling based optimal control to vastly reduce the compute time for easy adoption and adaptation to real time applications such as on-road autonomous driving in the presence of dynamic actors. Presented with training examples, a spatio-temporal CNN learns to predict the optimal mean control over a finite horizon that precludes further resampling, an iterative process that makes sampling based optimal control formulations difficult to adopt in real time settings. Generating control samples around the network-predicted optimal mean retains the advantage of sample diversity while enabling real time rollout of trajectories that seamlessly avoids multiple dynamic obstacles in an on-road navigation setting. Further the 3D convolution network architecture implicitly learns the future trajectories of the dynamic agents in the scene resulting in successful collision free navigation despite no explicit future trajectory prediction. We show performance gain over multiple baselines in a number of on-road scenes through closed loop simulations in CARLA. We also showcase the real world applicability of our system by running it on our custom Autonomous Driving Platform (AutoDP).