Arash K. Ushani

aushani@gmail.com

I am currently with Boston Dynamics as a Senior Staff Machine Learning Engineer. I work on problems in perception and machine learning for warehouse robots.

I completed my PhD in the Computer Science and Engineering department at the University of Michigan, Ann Arbor. I worked with Professor Ryan Eustice as a member of the Perceptual Robotics Laboratory. My research focused on object tracking and dynamic motion estimation for mobile robotics, specifically for autonomous vehicle applications.

Prior to that, I performed my undergraduate studies at Olin College of Engineering and spent two years working at Liquid Robotics in Silicon Valley.

In my spare time, I like to dabble in astrophotography.

Publications

Thesis

Understanding a Dynamic World: Dynamic Motion Estimation for Autonomous Driving using LIDAR
Arash Ushani
PhD Thesis, Department of Computer Science and Engineering, University of Michigan, Ann Arbor, MI. August 2018.
[PDF] [BibTeX]

Journals

Boosting Shape Registration Algorithms via Reproducing Kernel Hilbert Space Regularizers
Steven A. Parkison, Maani Ghaffari, Lu Gan, Ray Zhang, Arash K. Ushani, and Ryan M. Eustice,
IEEE Robotics and Automation Letters, 2019
[PDF] [BibTeX]

University of Michigan North Campus Long-Term Vision and Lidar Dataset
Nicholas Carlevaris-Bianco, Arash K. Ushani, and Ryan M. Eustice,
International Journal of Robotics Research, 2016.
[PDF] [BibTeX]

Conferences and Workshops

Feature Learning for Scene Flow Estimation
Arash K. Ushani and Ryan M. Eustice,
In Proceedings of the Conference on Robot Learning, Zurich, Switzerland, October 2018.
[PDF] [BibTeX] [Code] [Video]

A Learning Approach for Real-Time Temporal Scene Flow Estimation from LIDAR Data
Arash K. Ushani, Ryan W. Wolcott, Jeffrey M. Walls, and Ryan M. Eustice,
In Proceedings of the IEEE/RSJ International Conference on Robotics and Automation, Singapore, May 2017.
[PDF] [BibTeX] [Slides] [Code] [Video]

A Learning Approach for Real-Time Temporal Scene Flow Estimation from LIDAR Data
Invited Talk, Midwest Robotics Workshop, May 2017.

Dynamic Obstacle Tracking
Arash K. Ushani, Ryan W. Wolcott, Jeffrey M. Walls, and Ryan M. Eustice,
Poster, Midwest Robotics Workshop, May 2016.

Continuous-Time Estimation for Dynamic Obstacle Tracking
Arash K. Ushani, Nicholas Carlevaris-Bianco, Alexander G. Cunningham, Enric Galceran and Ryan M. Eustice,
In Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems, Hamburg, Germany, September 2015.
[PDF] [BibTeX]

Human Reach-to-Grasp Compensation with Object Pose Uncertainty
Qiushi Fu, Arash Ushani, Leif Jentoft, Robert D. Howe, and Marco Santello,
35th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, Osaka, Japan, July 2013.
[PDF] [BibTeX]

Projects

Warehouse Robotics

Next Generation Vehicle

The Next Generation Vehicle (NGV) project was a collaboration between the University of Michigan and Ford Motor Company to create a self driving car. Each vehicle is instrumented with sensors such as Velodyne LIDAR scanners, radar, and cameras and is capable of driving autonomously. I built algorithms to help detect and track dynamic obstacles on this platform.

Teaching

Computer Vision

I was a Graduate Student Instructor for a senior-level course in computer vision during fall 2015 at the University of Michigan, taught by Professor Matthew Johnson-Roberson. I taught a weekly discussion section and held office hours.

Discussion 1: [Slides] [MATLAB Exercise]
Discussion 2: [Slides] [MATLAB Exercise] [DLT Exercise]
Discussion 3: [Slides]
Discussion 4: [Slides]
Discussion 5: [Slides] [MATLAB Exercise]
Discussion 6: [Slides] [MATLAB Exercise]
Discussion 7: [Slides] [MATLAB Exercise]
Discussion 8: [Slides] [MATLAB Exercise]

Modeling and Control

At Olin College, for three years I was a teaching assistant. I taught a freshman-level class that introduced students to concepts in modeling and control, mainly focused on applications with electronic circuits. I taught two weekly tutorial sessions, held office hours, and graded labs.