| Seznam |
| Téma: | Plánování pohybu s omezeními |
|---|---|
| Katedra: | Katedra kybernetiky |
| Vedoucí: | Ing. Vojtěch Vonásek, Ph.D. |
| Vypsáno jako: | Diplomová práce, Bakalářská práce, Semestrální projekt |
| Popis: | Generally, motion planning aims to find a collision-free path for an object (robot) amongst obstacles. Sampling-based methods solve the problem by a randomized sampling of the related configuration space. In many applications, motions of the robots are further constrained, e.g., due to system kinematics, dynamics, or task constraints. The constraints induce a low-dimensional manifold that contains the solution. Usually, an analytic description of the manifold is not known, which makes it challenging for random sampling.
This project aims to implement a variant of Rapidly-exploring Random Tree (RRT) for the planning under constraints, namely the Atlas-RRT method. We assume a task for a 3D object moving in 3D space. The output of the project is a working implementation of the Atlas-RRT, scripts for visualization of the outputs (e.g., using Blender or similar tools), and a short technical report describing implemented methods and experimental results. The advisor will provide the benchmark for the verification of the method. Requirements: good knowledge of c/c++ and Python. Good knowledge of 3D modeling tools (Blender, Maya, Meshlab, etc.) is welcome. |