ICAPS 2018

TU Delft

June 24-29, 2018

ICAPS Conference

  • Medium Size (Usually 2 tracks going on)
  • High (but \(<\infty\)) PDDL:Deep Neural Network Ratio
  • Lots of provided food
  • Good doctoral consortium
  • Delft: A Dutch college town

An On-Line Planner for POMDPs with Large

Discrete Action Space: A Quantile-Based Approach

Main Idea:

Use a cross entropy-like method to choose which action nodes to explore.

Erli Wang, Hanna Kurniawati, Dirk Kroese

Effective Footstep Planning for Humanoids Using Homotopy-Class Guidance

Vinitha Ranganeni, Oren Salzman, Maxim Likhachev

Main Idea:

Use user-defined homotopy paths to create good heuristics for A*

 

Lazy Receding Horizon A* for Efficient Path Planning

in Graphs with Expensive-to-Evaluate Edges

Aditya Mandalika, Oren Salzman, Siddhartha Srinivasa

Main Idea:

Balance edge evaluations and graph operations for minimum run time

Amsterdam

Dutch Speeding Ticket

Wittenberg

Fort Loncin, Belgium

Berlin

Witnica, Poland

https://www.facebook.com/dariuszedward.jaworski/videos/2025891547482397/

ICAPS 2018

By Zachary Sunberg

ICAPS 2018

  • 483