Welcome to Multi-agent Multi-contact Path Planning.
This demonstration presents the first attempt using HPP, a new path planner, for multi-agent multi-contact path planning tasks. More specifically, this project deals with the case where there are a few HyQ robots on a Boeing 747-400 airplane. Each robot needs to move to the emergency exit as specified. We first generate the trajectory, based on which the contact are specified. Finally we illustrate the locomotion in animation. This work is a also a first attempt introducing Agent-Oriented concepts to multi-contact path planning.
The Robot, The Airplane and The Platform
This project chose Boeing 747-400 as the airplane and simplified the details in Blender before loading to HPP as the environment. This model was chosen because it has two emergency exits leading to the wings, which gives nicer visual effect. HyQ robots were used because not all the authors were fully granted to use some other robots. This implementation made use of the Python API of HPP.
Trajectory Generation
You first generate the trajectory as in the video
Contact Generation
You first generate the contact sequence for each agent
Authors and Contributors
This project was a first attempt using HPP for multi-agent path planning with consideration of contacts. The project unveiled possible changes of HPP for better hosting multiple agents. The current HPP implementation can not host multiple problems in parallel. Reuse of intermediate planning results are possible but not implemented. The project implemented a naive collision avoidance algorithm, which could be further optimised. The project was a part of the internship of Shuai Wang (a.k.a Robert White) supervised by Dr. Nicolas Mansard (@nmansard). Detailed help and guidance from Dr. Steve Tonneau (@stonneau), Mr. Joseph Mirabel (@jmirabel) and colleauges were also much appreciated.
Support or Contact
Please contact Robert for any questions or comments.