Jason Yosinski

About Me

Hi, I'm a Ph.D. student in Computer Science at the Cornell Creative Machines Lab, where I research artificial intelligence, robotics, and machine learning. Recent projects:

Nevermind all this, just show me all the videos!

Recent Papers more »

1. Sara Lohmann, Jason Yosinski, Eric Gold, Jeff Clune, Jeremy Blum and Hod Lipson. “Aracna: An Open-Source Quadruped Platform for Evolutionary Robotics” Proceedings of the 13th International Conference on the Synthesis and Simulation of Living Systems (to appear). 19 July 2012.
   pdf | abstract▾ | bib▾

   We describe a new, quadruped robot platform, Aracna, which requires non-intuitive motor commands in order to locomote and thus provides an interesting challenge for gait learning algorithms, such as those frequently developed in the Evolutionary Computation and Artificial Life communities. Aracna is an open-source hardware project composed of off-the-shelf and 3D-printed parts, enabling other research teams to modify its design according to their scientific needs. Aracna was designed to overcome the shortcomings of a previous quadruped robot platform, whose legs were so heavy that the motors could not reliably execute the commands sent to them. We avoid this problem by locating all motors in the body core instead of on the legs and through a design which enables the servos to have a greater mechanical advantage. Specifically, each of the four legs has two joints controlled by separate four-bar linkage mechanisms that drive the pitch of the hip joint and knee joint. This novel design causes unconventional kinematics, creating an opportunity for gait-learning algorithms, which excel in counter-intuitive design spaces where human engineers tend to underperform. Because it is low-cost, flexible, kinematically interesting, and and improvement over a previous design, Aracna provides a useful new hardware platform for testing algorithms that automatically generate robotic behaviors.

   @inproceedings{lohmann2012aracna:-an-open-source-quadruped,
   	Author = {Lohmann, Sara and Yosinski, Jason and Gold, Eric and Clune, Jeff and Blum, Jeremy and Lipson, Hod},
   	Booktitle = {Proceedings of the 13th International Conference on the Synthesis and Simulation of Living Systems},
   	Title = {Aracna: An Open-Source Quadruped Platform for Evolutionary Robotics},
   	Year = {2012 (to appear)},
   }
   
   

2. Jason Yosinski and Hod Lipson. “Visually Debugging Restricted Boltzmann Machine Training with a 3D Example” Presented at Representation Learning Workshop, 29th International Conference on Machine Learning. 1 July 2012.
   pdf | abstract▾ | bib▾

   Restricted Boltzmann Machines (RBMs) are being applied to a growing number of problems with great success. In the process of training an RBM one must pick a number of parameters, but often these parameters are brittle and produce poor performance when slightly off. Here we describe several useful visualizations to assist in choosing appropriate values for these parameters. We also demonstrate a successful application of an RBM to a unique domain: learning a representation of synthetic 3D shapes.

   @inproceedings{yosinski2012visually-debugging-restricted,
   	Author = {Jason Yosinski and Hod Lipson},
   	Booktitle = {Presented at Representation Learning Workshop, 29th International Conference on Machine Learning},
   	Title = {Visually Debugging Restricted Boltzmann Machine Training with a 3D Example},
   	Year = {2012}
   }
   
   

3. Jason Yosinski, Jeff Clune, Diana Hidalgo, Sarah Nguyen, Juan Cristobal Zagal, and Hod Lipson. “Evolving Robot Gaits in Hardware: the HyperNEAT Generative Encoding Vs. Parameter Optimization” Proceedings of the 20th European Conference on Artificial Life, Paris, France. pp 890-897. 8 August 2011.
   pdf | abstract▾ | bib▾

   Creating gaits for legged robots is an important task to enable robots to access rugged terrain, yet designing such gaits by hand is a challenging and time-consuming process. In this paper we investigate various algorithms for automating the creation of quadruped gaits. Because many robots do not have accurate simulators, we test gait-learning algorithms entirely on a physical robot. We compare the performance of two classes of gait-learning algorithms: locally searching parameterized motion models and evolving artificial neural networks with the HyperNEAT generative encoding. Specifically, we test six different parameterized learning strategies: uniform and Gaussian random hill climbing, policy gradient reinforcement learning, Nelder-Mead simplex, a random baseline, and a new method that builds a model of the fitness landscape with linear regression to guide further exploration. While all parameter search methods outperform a manually-designed gait, only the linear regression and Nelder-Mead simplex strategies outperform a random baseline strategy. Gaits evolved with HyperNEAT perform considerably better than all parameterized local search methods and produce gaits nearly 9 times faster than a hand-designed gait. The best HyperNEAT gaits exhibit complex motion patterns that contain multiple frequencies, yet are regular in that the leg movements are coordinated.

   @InProceedings{Yosinski2011EvolvedGaits,
     author =       {Jason Yosinski and Jeff Clune and Diana Hidalgo and Sarah Nguyen and Juan Cristobal Zagal and Hod Lipson},
     title =        {Evolving Robot Gaits in Hardware: the HyperNEAT Generative Encoding Vs. Parameter Optimization},
     booktitle =    {Proceedings of the 20th European Conference on Artificial Life},
     year =         {2011},
     month =        {August},
     numpages =     {8},
     location =     {Paris, France},
   }
   

4. Jason Yosinski and Cooper Bills. “MAV Stabilization using Machine Learning and Onboard Sensors” Technical Report CS6780, Cornell University (Posted to ArXiv 20 Feb 2012). 10 December 2010.
   pdf | abstract▾ | bib▾

   In many situations, Miniature Aerial Vehicles (MAVs) are limited to using only on-board sensors for navigation. This limits the data available to algorithms used for stabilization and localization, and current control methods are often insufficient to allow reliable hovering in place or trajectory following. In this research, we explore using machine learning to predict the drift (flight path errors) of an MAV while executing a desired flight path. This predicted drift will allow the MAV to adjust it’s flightpath to maintain a desired course.

   @techreport{yosinski2010mav,
     title={MAV Stabilization using Machine Learning and Onboard Sensors},
     author={Yosinski, J. and Bills, C.},
     year={2010},
     institution={Technical Report CS6780, Cornell University}
   }
   
   

more »

Personal

• My calendar, maybe useful for scheduling meetings.
• Slides from Machine Learning Summer School, in Kyoto, Japan, 2012.
• A bit of science fiction I wrote for a class.
• Fall Break 2011 trip to Lake George with other grad students.
• Flood cleanup in Owego, NY.
• Five week Europe and San Francisco conference trip during the summer of 2011.
• Fireworks on the First of July, because Ithaca is weird.
• Backpacking trip in Ithaca.
• Cross-country skiing for the first time, in Ithaca.
• Trip to Vermont with my roommates.
• See some pictures from my summer 2010 backpacking trip through South America.
• The Robotics Program at Eliot Middle School, created by Steve Crosby and I (now Stem X, run by Erik Dreyer).
• Find me on Twitter, Google Plus, Facebook, and Github.