Service to the planning community
Over the years, I’ve made some community service contributions to the AI and planning community.
Cognitive Robotics Summer School (CRSS) 2017
For a week in June 2017, my lab hosted the Cognitive Robotics Summer School (CRSS), an event held prior to the ICAPS 2017 conference while many researchers were on the east coast. I participated in helping organize this summer school during the event and for several months before.
My contributions to the summer school included:
-
Designing half of lab 3, which entailed a modeling exercise for students. Specifically, students were tasked with modeling a (simplified version!) of the starship scenario noted below from the ICKEPS competition.
-
Helping design and implement lab 4, which entailed asking the students to experiment with different motion planners for robotic manipulation. I helped set up a live demo featuring our lab’s WAM arm, in which the arm would continually replan to achieve a certain stack of blocks. If a person came over and moved the blocks around to arbitrary locations, the robot would notice and automatically come up with a new plan to re-stack in the proper configuration. This involved integrating my research algorithm (Pike) with a generative planner, our arm’s controller and motion planner, and a state estimation module.
-
Helping design and implement the grand challenge, the final assignment for the students. I was responsible for integrating the above-mentioned robotic arm into the final scenario. The students were able to model scenarios which, when executed in hardware, would cause the robotic arm to pick up a quadcopter, place it on a second robot (which would drive the quadcopter around), and then the quadcopter would fly around to achieve a mission.
Here is a picture from the grand challenge, in which the WAM arm picks up one of two quadcopters, and is in the midst of placing it down onto a small mobile robot, before it can drive off and launch the quadcopter in a different area:
ICKEPS 2016 Competition
During the summer of 2016, I contributed to the fifth International Competition on Knowledge Engineering for Planning and Scheduling (ICKEPS 2016), a competition that was part of the ICAPS 2016 conference.
Specifically, collaborating with Tiago Vaquero, we contributed a starship-themed exploration scenario for participants to model. The scenario encompassed complex temporal constraints, preconditions and effects of actions, resource requirements, and more.
To see the scenario and others: Scenarios
Temporal Plan Visualizer
In collaboration with Christian Muise, the developer of planning.domains, I developed a plugin that allows modelers to visualize the outputs of a temporal planner.
planning.domains is a collection of tools for those in the planning community; it encompases (1) a curated archive of planning domains and problems, along with an API to access them programatically, (2) an automated planner in the cloud for finding solutions to those planning problems, and (3) a web-based editor for modeling new problems. It is in this third area that I made my contribution.
The online editor, editor.planning.domains is designed to have an extensible plugin architecture. As such, I wrote a Javascript plugin that parses the output of a temporal planner (such as optic
) and displays a timeline of actions to the user.