ATS-RRT: an improved RRT algorithm based on alternative paths and triangular area sampling

The Rapidly Exploring Random Tree Star (RRT*) is a probabilistically complete algorithm. It is recognized as a better path planning algorithm, but its path quality and path planning speed still have room for improvement. This paper proposes an improved RRT* algorithm based on alternative paths and t...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Advanced robotics Ročník 37; číslo 10; s. 605 - 620
Hlavní autoři: Zhang, Zhi-wei, Jia, Yun-wei, Su, Qi-qi, Chen, Xiao-tong, Fu, Bang-peng
Médium: Journal Article
Jazyk:angličtina
Vydáno: Taylor & Francis 19.05.2023
Témata:
direct connection strategy
path planning
RRT
sampling-based algorithms
ISSN:0169-1864, 1568-5535
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:The Rapidly Exploring Random Tree Star (RRT*) is a probabilistically complete algorithm. It is recognized as a better path planning algorithm, but its path quality and path planning speed still have room for improvement. This paper proposes an improved RRT* algorithm based on alternative paths and triangular area sampling (ATS-RRT*). The alternative paths strategy generates multiple initial paths based on whether the sample points can communicate with the target points and set the path with the smallest cost as the final initial path, which can speed up the initial path planning and improve the initial path finding rate. The triangular area sampling strategy combines every three adjacent nodes to generate some triangle areas and corresponding half-triangle areas. The path quality can be improved quickly by limiting the sampling in these triangle areas. In addition, the direct connection strategy with triangle nodes and the tabu table using in the Rewire process also speeds up the algorithm. Experiments show that the speed of finding the initial path and the success rate of finding the suboptimal path are improved by 2.3 and 1.45 times respectively compared with RRT*, Quick-RRT*, and Informed + Quick-RRT*.
ISSN:0169-1864
1568-5535
DOI:10.1080/01691864.2023.2174817