OSS4DM 2017

We are proud to announce a new workshop for Open Source Software for Decision Making, the scope of which includes reinforcement learning, planning and learning in fully- and partially-observable Markov decision processes, decision system certification and validation, advanced algorithms and data structures for decision making, and human-robot interface languages. Special emphasis is placed on open-sourced software and open research - tools which are openly accessible for use in industry and academia alike.

The decision systems used in academia and industry are often proprietary, and open-sourced software suites such as BURLAP, APPL, and MADP are neither well-known, nor widely circulated. These open-source software projects would benefit from broader community involvement and coordination. Furthermore, while there are high quality datasets and machine learning packages such as scikit-learn, dlib, and Weka, these do not address most decision making problems, such as the control of dynamic systems under uncertainty.

Sponsored by the Artificial Intelligence Journal.

Presentations

01: Scott Sanner
RDDL - the Relational Dynamic Influence Diagram Language

02: Louise Poubel
Gazebo - robot simulation made easy

03: Max Egorov
POMDPs.jl - a unified POMDP framework

04: Chelsea Finn
Guided Policy Search - complete algorithm and robot interface

05: Rocky Duan
RLLAB - a reinforcement learning framework

06: David Silver
The Udacity Open-Sourced Self Driving Car

07: Benoit Landry
MotionPlanning.jl - A Julia package for robotic motion planning

08: Benoit Landry
Drake - planning and contorl for nonlinear dynamical systems

09: Zach Sunberg
POMDPs.jl - challenges and lessons learned

10: Peter Karkus
Approximate POMDP Solving Software

11: Jim (Linxi) Fan
World of Bits

Schedule

Event Time Description
Registration Opens 8:30 AM
Workshop Opening 9:00 AM Prof. Mykel Kochenderfer from Stanford
         RDDL - the Relational Dynamic Influence Diagram Language 9:20 AM Prof. Scott Sanner from U. Toronto
         Gazebo - robot simulation made easy 9:45 AM Louise Poubel from Gazebo
         POMDPs.jl - a unified POMDP framework 10:10 AM Max Egorov from Stanford
Break (20 min)
         Guided Policy Search - complete algorithm and robot interface 10:50 AM Chelsea Finn from UC Berkeley
         RLLAB - a reinforcement learning framework 11:10 AM Rocky Duan from UC Berkeley
         The Udacity Open-Sourced Self-Driving Car 11:30 AM David Silver from Udacity
Lunch Break (70 min)
         MotionPlanning.jl - A Julia package for robotic motion planning 1:00 PM Benoit Landry from Stanford
         Drake - planning and control for nonlinear dynamical systems 1:20 PM Benoit Landry from Stanford
         POMDPs.jl - challenges and lessons learned 1:40 PM Zach Sunberg from Stanford
Break (20 min)
         Approximate POMDP Solving Software 2:20 PM Peter Karkus from Natl. U. of Singapore
         World of Bits 2:40 PM Jim (Linxi) Fan from Stanford
         The State of OSS4DM and Closing Comments 3:00 PM Tim Wheeler from Stanford
Fin 3:30 PM

Registration

Registration is now closed. Stay tuned for next time!

Directions

Tresidder Union, 459 Lagunita Dr, Stanford, CA 94305, USA Parking is available directly at the venue. You can pay with cash or credit card at the kiosk. Display your parking slip on your dashboard.

  • From 280: Exit Sand Hill Road east toward Stanford. Continue east, turning right at the traffic light on Santa Cruz Avenue. Make an immediate left onto Junipero Serra Boulevard. Turn left on Campus Drive East then make a left turn onto Mayfield Avenue.
  • From 101: Take the Embarcadero Road exit west toward Stanford. At El Camino Real, Embarcadero turns into Galvez Street as it enters the university. Stay in the left lane and continue toward the center of campus. Make a left turn at Campus Drive East. Follow the signs to the Stanford Bookstore. Turn right on Mayfield Avenue.
  • From El Camino Real: Exit West at University Avenue, then toward the hills (away from the center of Palo Alto). As you enter Stanford, University Avenue becomes Palm Drive. Go through one traffic light, and turn left onto Campus Drive. Campus Drive will turn into Campus Drive East. Follow the signs to the Stanford Bookstore. Turn right on Mayfield Avenue.

Dining

Lunch will not be provided at the workshop, but there will be coffee and pastries available at opening and for the breaks throughout the day.

There are many eateries within walking distance for lunch. Russo Cafe is right next to Paul Brest Hall but may get crowded at lunch. There are many other options at Tresidder Marketplace including: Treehouse, CoHo, Subway, Panda Express, Starbucks, and Jamba Juice.

Wifi

Please use the "Stanford Visitor" network available across campus.

Contact

Symposium chairs:

Local arrangements:

If you are interested in sponsoring this event, please contact Tim Wheeler.