Robert Pěnička presents Enhancing sampling-based planning with a library of paths
On 2026-03-31 11:00:00 at E112, Karlovo náměstí 13, Praha 2
In the upcoming seminar, we will present our recently published approach to one
of the most persistent challenges in robotics: path planning for 3D solid
objects in environments with narrow passages. While traditional sampling-based
planners often struggle with low sampling probabilities in confined spaces and
typically start each task from scratch, the presented research introduces
RRT-LIB, a method that leverages a library of past solutions to guide future
planning. By utilizing 3D shape similarity and the Iterative Closest Point
(ICP)
algorithm, the system can identify previously seen objects similar to a new
query and adapt their historical trajectories as "guide paths." This
experience-driven strategy significantly reduces planning time—by up to 85%
in
some scenarios—allowing for more efficient autonomous navigation and
assembly.
The seminar will be held in a short format of 20 minutes + 10 minutes Q&A.
Reference:
Michal Minařík, Vojtěch Vonásek, Robert Pěnička, "Enhancing
sampling-based
planning with a library of paths," in Robotics and Autonomous Systems, vol.
198,
2026, doi: 10.1016/j.robot.2026.105334.
Paper: https://www.sciencedirect.com/science/article/pii/S0921889026000072
Video: https://youtu.be/1BTlWC742Aw?si=YnZRcAFjKunlE3fK
Code: https://github.com/m-minarik/rrtlib
of the most persistent challenges in robotics: path planning for 3D solid
objects in environments with narrow passages. While traditional sampling-based
planners often struggle with low sampling probabilities in confined spaces and
typically start each task from scratch, the presented research introduces
RRT-LIB, a method that leverages a library of past solutions to guide future
planning. By utilizing 3D shape similarity and the Iterative Closest Point
(ICP)
algorithm, the system can identify previously seen objects similar to a new
query and adapt their historical trajectories as "guide paths." This
experience-driven strategy significantly reduces planning time—by up to 85%
in
some scenarios—allowing for more efficient autonomous navigation and
assembly.
The seminar will be held in a short format of 20 minutes + 10 minutes Q&A.
Reference:
Michal Minařík, Vojtěch Vonásek, Robert Pěnička, "Enhancing
sampling-based
planning with a library of paths," in Robotics and Autonomous Systems, vol.
198,
2026, doi: 10.1016/j.robot.2026.105334.
Paper: https://www.sciencedirect.com/science/article/pii/S0921889026000072
Video: https://youtu.be/1BTlWC742Aw?si=YnZRcAFjKunlE3fK
Code: https://github.com/m-minarik/rrtlib