Collision-Free Multi Robot Trajectory Optimization in Unknown\n Environments using Decentralized Trajectory Planning

Multi robot systems have the potential to be utilized in a variety of\napplications. In most of the previous works, the trajectory generation for\nmulti robot systems is implemented in known environments. To overcome that we\npresent an online trajectory optimization algorithm that utilizes communication\nof robots' current states to account to the other robots while using local\nobject based maps for identifying obstacles. Based upon this data, we predict\nthe trajectory expected to be traversed by the robots and utilize that to avoid\ncollisions by formulating regions of free space that the robot can be without\ncolliding with other robots and obstacles. A trajectory is optimized\nconstraining the robot to remain within this region.The proposed method is\ntested in simulations on Gazebo using ROS.\n