The paper introduces CBFKIT, an open-source Python/ROS toolbox for safe robotics planning and control under uncertainty. The key highlights are:
CBFKIT provides a modular and extendable framework for designing control barrier functions (CBFs) to guarantee safety and forward invariance of robotic systems. It supports a variety of control-affine models, including deterministic ODEs, ODEs with bounded disturbances, and stochastic differential equations.
The toolbox utilizes functional programming principles to ensure code reliability, maintainability, and ease of debugging. It offers template functions for dynamics, controllers, estimators, and other key components, allowing users to easily instantiate and compose them.
CBFKIT integrates with the Robot Operating System (ROS), enabling users to connect the CBF-based control framework with ROS-enabled robotic platforms. This allows for the setup of multi-robot applications, encoding of environments and maps, and integration with predictive motion planning algorithms.
The toolbox provides multiple CBF variations and algorithms for robot control, including quadratic program-based approaches for deterministic and stochastic systems. It also includes tutorials and examples demonstrating the use of CBFKIT for various robotic systems, such as the Toyota Human Support Robot (HSR).
The authors demonstrate the application of CBFKIT on the Toyota HSR robot, both in simulation and physical experiments, where the CBF-based controller successfully navigates the robot to a goal location while avoiding collisions with a human agent.
toiselle kielelle
lähdeaineistosta
arxiv.org
Tärkeimmät oivallukset
by Mitchell Bla... klo arxiv.org 04-11-2024
https://arxiv.org/pdf/2404.07158.pdfSyvällisempiä Kysymyksiä