Multi-agent Trajectory Optimization

Description

We have focused on understanding the novel trajectory planning technique proposed in the reference paper- "Efficient trajectory planning for multiple non-holonomic mobile robots via prioritized trajectory optimization". Further, we have analyzed and discussed the methodology of the solution. The simulation results are replicated and the significant inferences stated by the authors are validated. Finally, the hyper-parameters of the solution are altered to extract additional inferences.

1. Software Requirements

• Ubuntu 20.04
• ROS Noetic
• Octomap
• Ipopt

2. Installation instructions

- Install Ipopt solver

Follow Ipopt Installation.

- Install dependencies

sudo apt-get install ros-noetic-octomap*
sudo apt-get install ros-noetic-dynamic-edt-3d
sudo apt-get install cppad

- Build:

Copy the folder files/multi_agent_opt_traj and paste it in the src folder of your catkin workspace. The following commands can then be run to build the package.

cd <catkin_ws> && catkin_make
source ~/catkin_ws/devel/setup.bash

3. Run Simulations

Warehouse

roslaunch multi_robot_traj_planner prioritized_plan_warehouse.launch 

Environment with random obstacles

roslaunch multi_robot_traj_planner prioritized_plan_random_env.launch

4. Results

1. Video recordings from our experiment are in the output/videos folder.
2. Adaptation results can be viewed in the output/inferences_* folders.

Simulation Data

Data gathered from the simulations can be viewed in the simulation_data folder.

4. Acknowledgments

Our implementation is built on top of the work of the official repo of the reference paperEfficient trajectory planning for multiple non-holonomic mobile robots via prioritized trajectory optimization