Joaquim Ortiz-Haro,Wolfgang Hönig,Valentin N. Hartmann,Marc Toussaint,Ludovic Righetti
Abstract:Rapidly-exploring Random Trees (RRT) and its variations have emerged as a robust and efficient tool for finding collision-free paths in robotic systems. However, adding dynamic constraints makes the motion planning problem significantly harder, as it requires solving two-value boundary problems (computationally expensive) or propagating random control inputs (uninformative). Alternatively, Iterative Discontinuity Bounded A* (iDb-A*), introduced in our previous study, combines search and optimization iteratively. The search step connects short trajectories (motion primitives) while allowing a bounded discontinuity between the motion primitives, which is later repaired in the trajectory optimization step.
What problem does this paper attempt to address?
The paper proposes a new motion planning algorithm called iDb-RRT for solving robot motion planning problems with dynamic constraints. In the traditional Rapidly-exploring Random Trees (RRT) method, adding dynamic constraints significantly increases the computational difficulty, as it requires solving boundary value problems or propagating random control inputs, which is not ideal in terms of efficiency and informativeness. However, iDb-RRT combines motion primitives and trajectory optimization to work within the RRT framework, by allowing limited discontinuity between motion primitives and subsequently repairing them through trajectory optimization, thus improving efficiency.
The iDb-RRT algorithm has the following characteristics:
1. It is probabilistically complete and can be implemented in either one-way or bi-directional mode.
2. By iteratively increasing the number of motion primitives and reducing the discontinuity bounds, the algorithm is able to find solutions.
3. In a benchmark test of 30 problems covering different systems, iDb-RRT was 10 times faster than existing methods in terms of finding solutions, particularly in complex scenarios requiring long trajectories or traversal through narrow passages.
Compared to other methods such as search-based, optimization-based, and sampling-based methods, the advantages of iDb-RRT are that it does not require solving expensive two-point boundary value problems and does not need to discretize the workspace. It utilizes pre-computed motion primitives and allows for discontinuity in the planning phase, which is then corrected through trajectory optimization, achieving probabilistic completeness.
Experimental results show that iDb-RRT outperforms existing algorithms based on sampling, search, and optimization in terms of computation time and solution cost, especially in high-dimensional, nonlinear, and long trajectory planning problems.