I am a Ph.D. candidate in Computer Science at Rutgers University. I am part of the PRACSYS lab, advised by Kostas Bekris. I am broadly interested in applying machine learning algorithms to improve the efficiency and robustness of robotic planning. I am excited about building robust, real-world robotics systems that integrate perception, planning and learning.
Prior to my Ph.D., I earned an MS in CS from Rutgers. I earned my undergraduate degree in Computer Science & Engineering from Amrita Vishwa Vidyapeetham University. In Summer 2018, I was an intern at Preferred Networks in Tokyo working with Wilson Ko and Kuniyuki Takahashi. In Summer 2022, I was a Research Intern at Bosch Austin working with Alessandro Allievi, Jarrett Holtz and Xuesu Xiao.
E-mail: aravind.siva@rutgers.edu
This page is perpetually under construction until I say otherwise. You can find me on Github and Twitter. My CV can be found here.
[Current Research] [Papers] [Teaching] [Projects] [Talks] [Courses] [Misc]Current Research
Currently, I am working on improving the efficiency of motion planning for robots with significant dynamics using data-driven techniques.
I also collaborate with the DATA-INSPIRE TRIPODS Institute on data-informed dynamical systems theory, with an emphasis on robotic control.
I have worked with our lab's KUKA LBR iiwa and RoboMantis robots.
Conference and Journal Papers
- Ewerton R. Vieira, Aravind Sivaramakrishnan, Yao Song, Edgar Granados, Marcio Gameiro, Konstantin Mischaikow, Ying Hung, Kostas E. Bekris. Data-Efficient Characterization of the Global Dynamics of Robot Controllers with Confidence Guarantees. Proceedings of the 2023 IEEE International Conference on Robotics and Automation (ICRA), 2023. [pdf][arXiv]
- Troy McMahon*, Aravind Sivaramakrishnan*, Edgar Granados and Kostas E. Bekris, A Survey on the Integration of Machine Learning with Sampling-based Motion Planning. Foundations and Trends® in Robotics: Vol. 9: No. 4, pp 266-327. [pdf] [arXiv] [webpage]
- Troy McMahon, Aravind Sivaramakrishnan, Kushal Kedia, Edgar Granados, Kostas E. Bekris, Terrain-Aware Learned Controllers for Kinodynamic Planning over Physically Simulated Terrains. Proceedings of the 2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2022. [pdf]
- Ewerton R. Vieira, Edgar Granados, Aravind Sivaramakrishnan, Marcio Gameiro, Konstantin Mischaikow, Kostas E. Bekris. Morse Graphs: Topological Tools for Analyzing the Global Dynamics of Robot Controllers. Proceedings of the 15th International Workshop on the Algorithmic Foundations of Robotics (WAFR), 2022. [pdf] [arXiv]
- Aravind Sivaramakrishnan, Edgar Granados, Seth Karten, Troy McMahon, Kostas E. Bekris. Improving Kinodynamic Planners for Vehicular Navigation with Learned Goal-Reaching Controllers. Proceedings of the 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2021. [pdf] [arXiv]
Workshop Papers
- Edgar Granados*, Aravind Sivaramakrishnan*, Troy McMahon, Zakary Littlefield, Kostas E. Bekris. ML4KP: a Light and Flexible Library for Development and Benchmarking of Sampling-Based Kinodynamic Planners. Machine Learning for Motion Planning workshop at ICRA 2021 (MLMP), 2021. [pdf]
- Seth Karten, Aravind Sivaramakrishnan, Edgar Granados, Troy McMahon, Kostas E. Bekris. Data-Efficient Learning of High-Quality Controls for Kinodynamic Planning used in Vehicular Navigation. Machine Learning for Motion Planning workshop at ICRA 2021 (MLMP), 2021. [pdf]
- Aravind Sivaramakrishnan, Zakary Littlefield, Kostas E. Bekris. Towards Learning Efficient Maneuver Sets for Kinodynamic Motion Planning. 7th ICAPS Workshop on Planning and Robotics (PlanRob), 2019. [pdf]
Other Papers
- Aravind Sivaramakrishnan, Madhusudhan Krishnamachari, Vidhya Balasubramanian. Recommending Customizable Products: A Multiple Choice Knapsack Solution. Proceedings of the 5th International Conference on Web Intelligence, Mining and Semantics (WIMS), 2015.[pdf] [code]
Teaching
In Spring 2023, I am the TA for CS590: Socially Cognizant Robotics.
I have instructed / TAed for CS440: Introduction to Artificial Intelligence (Su17, Sp18, Sp19, Su19, Fa19, Su20, Sp21, Su21, Sp22), CS520: Introduction to Artificial Intelligence (Sp17, Fa17, Fa19, Sp20, Fa20, Fa21, Fa22) and CS460/560: Introduction to Computational Robotics (Fa18).
Projects
- irl-lab
A WIP implementation of popular Inverse Reinforcement Learning algorithms for various tasks.
- Fake News Challenge
A better than baseline model written in Keras for the Fake News Challenge.
- Sampling Based Planners
A Python implementation of PRM and RRT for a simple 2D navigation task with polygonal obstacles.
- Leap Motion Data Recorder
A minimal C++ based solution for recording and playback of hand tracking data recorded using the Leap Motion controller.
- Deep Averaging Networks
A Keras implementation of the model described in this paper for factoid question answering.
- Vanilla GANs
A numpy implementation of fully connected Generative Adversarial Networks (GANs) on the MNIST dataset.
Talks
- Learning Efficient Maneuver Sets for Constrained Sampling-based Planning. 14th Annual New England Manipulation Symposium (NEMS), 2019. [poster]
- Some more that I haven't gotten around to listing...
Coursework
- CS513: Design & Analysis of Data Structures & Algorithms [syllabus]
- CS520: Introduction to Artificial Intelligence [syllabus]
- CS525: Brain-Inspired Computing [syllabus]
- CS535: Pattern Recognition [syllabus]
- CS536: Machine Learning [syllabus]
- CS596: Foundations of Computer & Data Science [syllabus]
- CS598: Computational Foundations of Robotics [syllabus]
- CS672: Algorithmic Robotics [syllabus]
- CS674: Mathematical Topics in AI & Optimization [syllabus]
Misc
- I attended the 3rd Summer School on Cognitive Robotics, held at USC in Los Angeles.
- I presented a tutorial on data-driven controllers for dynamical robotic systems as part of the TRIPODS Data Science Boot Camp: Dynamics, Data Analysis, and Robotic Control.
- I served as the Treasurer (2021-22) of the Computer Science Graduate Student Society (CSGSS).
- I will believe "AI is taking over the world!" hype the day it can understand quaternions. (the tweet I am quoting is from a Google Brain researcher in Robotics)