Long-Horizon Planning

Most tasks faced by a robot require more than a single “action”—for example, to cook a meal, there are many steps in a recipe; in order to navigate a building, a robot must explore many rooms, find feasible unblocked paths, open doors, and so on; construction tasks have many interlocking dependencies (e.g., peg A in hole B before screw C into socket D, etc.).

Critical to solving these kinds of tasks is some abstraction over the set of actions a robot can do, allowing methods to reason over what to do at a higher level (i.e., rather than thinking about joint angles throughout the entire motion, thinking at the level of “pick up object A” and “place object B on object C”). Designing or automatically finding suitable abstractions for a problem (do you assume to know the set of actions the robot can perform, or is this something you must find out?), finding ways of providing feedback through abstraction (e.g., what happens when you can’t find a motion to pick up an object? Do you never pick up that object again? Is another object blocking it, or is it something else? How do you discover and inform search about this?), and efficiently searching over the infinite combinatorial explosion of options is essential to solving these problems effectively.


2024

  1. Accelerating Long-Horizon Planning with Affordance-Directed Dynamic Grounding of Abstract Strategies
    In IEEE International Conference on Robotics and Automation

2023

  1. Solving Rearrangement Puzzles using Path Defragmentation in Factored State Spaces
    S. Bora Bayraktar , Andreas OrtheyZachary KingstonMarc Toussaint , and Lydia E. Kavraki
    IEEE Robotics and Automation Letters
  2. Object Reconfiguration with Simulation-Derived Feasible Actions
    Yiyuan Lee , Wil ThomasonZachary Kingston , and Lydia E. Kavraki
    In IEEE International Conference on Robotics and Automation
  3. Optimal Grasps and Placements for Task and Motion Planning in Clutter
    In IEEE International Conference on Robotics and Automation
  4. Scaling Multimodal Planning: Using Experience and Informing Discrete Search
    IEEE Transactions on Robotics

2021

  1. Using Experience to Improve Constrained Planning on Foliations for Multi-Modal Problems
    In IEEE/RSJ International Conference on Intelligent Robots and Systems
  2. Finite Horizon Synthesis for Probabilistic Manipulation Domains
    In IEEE International Conference on Robotics and Automation

2020

  1. Informing Multi-Modal Planning with Synergistic Discrete Leads
    In IEEE International Conference on Robotics and Automation

2018

  1. An Incremental Constraint-Based Framework for Task and Motion Planning
    The International Journal of Robotics Research, Special Issue on the 2016 Robotics: Science and Systems Conference

2016

  1. Incremental Task and Motion Planning: A Constraint-Based Approach
    In Robotics: Science and Systems