The project proposes to develop new algorithms for the kinodynamic planning problem that far surpass the state-of-the-art methods upon which a large fraction of autonomous systems are based. The proposed methods are based on two key ideas, the well-known and highly used Rapidly exploring Random Tree (RRT) algorithm, and bang-bang steering, which is a recent breakthrough that enables time-optimal trajectory segments to be computed orders of magnitude faster than earlier methods. The expected results are powerful new algorithms and software that take autonomy to the next level of speed and reliability, and an improved understanding of fundamental problems in robotics, computer science, sampling theory, and optimal control. This in turn benefits society through enabling the development of more advanced autonomous systems.

2024

2028