James Alan Preiss

I am a postdoctoral scholar in the Department of Computing + Mathematical Sciences at Caltech, where I am co-advised by Yisong Yue, Soon-Jo Chung, and Adam Wierman. I completed my Ph.D. with Prof. Gaurav Sukhatme at the USC Robotic Embedded Systems Lab (RESL).

I am interested in robotics, machine learning, control theory, optimization, and their intersections — both theoretical and practical. Recent projects include regret bounds for online controller selection, policy gradient reinforcement learning theory, combining deep dynamics models with traditional control, and developing the framework of suboptimal covering numbers to quantify the "size" of infinite sets of optimal control tasks.

Earlier projects include trajectory optimization hardness analysis, empirical projects in RL, multi-robot motion planning, and co-developing/maintaining the Crazyswarm platform for multi-quadrotor research with Wolfgang Hönig.

Before starting my Ph.D., I developed software for 3d scanning at Geomagic and interactive statistics at the JMP division of SAS Institute. I completed a B.S. in Mathematics and B.A. in Photography at The Evergreen State College in Olympia, WA.

This page last updated January 2023.

• Online switching control with stability and regret guarantees.
  Yingying Li, James A. Preiss, Na Li, Yiheng Lin, Adam Wierman, and Jeff Shamma.
  Learning for Dynamics & Control, 2023. [preprint]
  Oral presentation
• Parameter Estimation for Deformable Objects in Robotic Manipulation Tasks
  David Millard, James A. Preiss, Jernej Barbič, and Gaurav S. Sukhatme.
  International Symposium of Robotics Research (ISRR), 2022. [Springer page]
• Online Adaptive Controller Selection in Time-Varying Systems: No-Regret via Contractive Perturbations.
  Yiheng Lin, James A. Preiss, Emile Anand, Yingying Li, Yisong Yue, and Adam Wierman.
  arXiv preprint, 2022. [preprint]
• Tracking Fast Trajectories with a Deformable Object using a Learned Model.
  James A. Preiss*, David Millard*, Tao Yao*, and Gaurav S. Sukhatme
  IEEE International Conference on Robotics and Automation (ICRA), 2022. [IEEE page]
• Suboptimal coverings for continuous spaces of control tasks.
  James A. Preiss and Gaurav S. Sukhatme.
  Conference on Learning for Dynamics and Control (L4DC), 2021. [PMLR page]
• Resilient Coverage: Exploring the Local-to-Global Trade-off.
  Ragesh K. Ramachandran, Lifeng Zhou, James A. Preiss, and Gaurav S. Sukhatme.
  IEEE International Conference on Intelligent Robots and Systems (IROS), 2020. [preprint]
• A Closer Look at Reinforcement Learning for Neural Network Architecture Search.
  James A. Preiss, Eugen Hotaj, and Hanna Mazzawi.
  ICLR Workshop on Neural Architecture Search, 2020. [pdf]
  Selected for contributed talk
• Analyzing the Variance of Policy Gradient Estimators for the Linear-Quadratic Regulator.
  James A. Preiss, Sébastien M. R. Arnold, Chen-Yu Wei and Marius Kloft.
  NeurIPS Workshop on Optimization Foundations for Reinforcement Learning, 2019. [preprint]
• Estimating Metric Scale Visual Odometry from Videos using 3D Convolutional Networks.
  Alexander S. Koumis, James A. Preiss, and Gaurav S. Sukhatme.
  IEEE International Conference on Intelligent Robots and Systems (IROS), 2019. [preprint]
• Resilience by Reconfiguration: Exploiting Heterogeneity in Robot Teams.
  Ragesh K. Ramachandran, James A. Preiss, and Gaurav S. Sukhatme.
  IEEE International Conference on Intelligent Robots and Systems (IROS), 2019. [preprint]
• Sim-to-(Multi)-Real: Transfer of Low-Level Robust Control Policies to Multiple Quadrotors.
  Artem Molchanov, Tao Chen, Wolfgang Hönig, James A. Preiss, Nora Ayanian, and Gaurav S. Sukhatme.
  IEEE International Conference on Intelligent Robots and Systems (IROS), 2019. [preprint] [project webpage]
• Understanding the Variance of Policy Gradient Estimators in Reinforcement Learning.
  Sébastien M. R. Arnold, James A. Preiss, Chen-Yu Wei and Marius Kloft.
  Southern California Machine Learning Symposium, 2019. [preprint of newer version]
  Awarded best poster
• Learning a System-ID Embedding Space for Domain Specialization with Deep Reinforcement Learning.
  James A. Preiss, Karol Hausman, Gaurav S. Sukhatme.
  NeurIPS Workshop on Reinforcement Learning under Partial Observability, 2018. [pdf]
• Simultaneous Self-Calibration and Navigation using Trajectory Optimization.
  James A. Preiss, Karol Hausman, Gaurav S. Sukhatme, and Stephan Weiss.
  International Journal of Robotics Research (IJRR) RSS special issue (invitation-only), 2018. [journal page] [preprint]
• Trajectory Planning for Quadrotor Swarms.
  Wolfgang Hönig, James A. Preiss, T.K. Satish Kumar, Gaurav S. Sukhatme, and Nora Ayanian.
  IEEE Transactions on Robotics (T-RO) special issue on aerial swarm robotics, 2018. [preprint] [video]
• Trajectory Optimization for Self-Calibration and Navigation.
  James A. Preiss, Karol Hausman, Gaurav S. Sukhatme, and Stephan Weiss.
  Robotics: Science and Systems (RSS), 2017. [pdf]
• Downwash-Aware Trajectory Planning for Large Quadcopter Teams.
  James A. Preiss, Wolfgang Hönig, Nora Ayanian, and Gaurav S. Sukhatme
  IEEE International Conference on Intelligent Robots and Systems (IROS), 2017. [preprint] [video]
• Crazyswarm: A Large Nano-Quadcopter Swarm.
  James A. Preiss*, Wolfgang Hönig*, Gaurav S. Sukhatme, and Nora Ayanian.
  IEEE International Conference on Robotics and Automation (ICRA), 2017. [pdf] [video]
• Observability-Aware Trajectory Optimization for Self-Calibration with Application to UAVs.
  Karol Hausman, James A. Preiss, Gaurav S. Sukhatme, and Stephan Weiss.
  IEEE Robotics and Automation Letters (RA-L), 2017. [preprint] [video]

* = equal contributors.

Ph.D. Dissertation

Characterizing and Improving Robot Learning: A Control-Theoretic Perspective.
James A. Preiss.
University of Southern California, 2022. [pdf]

• crazyswarm - our full stack for flying 49+ Crazyflie quadcopters in a motion capture space. Includes object tracking, improved radio protocols, onboard sensor fusion, nonlinear controller, onboard trajectory planning, ROS node, and Python scripting.
• reproducible papers - using Makefiles to make a complete paper, including experiments, reproducible from a few shell commands.
• smoothener - convert multi-robot waypoint sequences into smooth piecewise polynomial trajectories. Implementation of the continuous portion of our IROS 2017 paper.
• cmath3d - a 3d math library for embedded C.
• pputils - utilities for working with piecewise polynomials in Matlab's ppform.
• quatcompress - compress a quaternion into 32 bits.

(first initial) (middle initial) (last name) @ caltech.edu