An abstract, efficient C++ library for autonomous high-level discrete task planning.
Imagine you have a robotic system equipped with motion planning, low-level control, and state estimation/observation. This robot can plan safe non-colliding continuous movements, execute these movements in real time, and may even be able to observe state feedback. Now put this robot in a kitchen, and tell it to efficiently and safely prepare meals for the customers. Your robot may know how to move from one location to the other, or plan a motion that flips a burger patty. But how does your robot know how to assemble each meal correctly, efficiently, and safely?
This task planning library gives you the tools and algorithms to utilize or design a discrete task planner, capable of satisfying complex Linear Temporal Logic (LTL) task specifications.
- Dijkstra's Algorithm (Single-objective shortest path)
- A* (Single-objective heuristic-guided shortest path)
- Bi-Objective A* (a.k.a. BOA*, Two-objective optimal trade-off Pareto Front)
- New Approach to Multi-Objective A* (a.k.a. NAMOA*, N-objective optimal trade-off Pareto Front)
The custom model generator and logic parser allow the user to create discrete Transition System models of an autonomous system. This library also includes simple model generators for a robotic manipulator capable of grasping and moving objects, and a grid-world mobile robot with rectangular regions of interest.
The Deterministic Task Planner will autonomously plan a sequence of actions that when executed from the initial state, satisfies a set of LTL tasks. The Bi-Objective Preference Planner and Multi-Objective Preference Planner will plan a sequence of actions that will satisfies a set of LTL tasks, while optimizing over multiple preference objectives.
- C++20
- Python3
- SpotLTL
- YAML-cpp
- Eigen
- Install Spot https://spot.lre.epita.fr/install.html and Eigen https://eigen.tuxfamily.org/dox/GettingStarted.html
git submodule update --init --recursivemkdir build && cd buildcmake ..make