In recent years, aerial swarm technology has developed rapidly. In order to accomplish a fully autonomous aerial swarm, a key technology is collaborative SLAM (CSLAM) for aerial swarms, which estimates the relative pose and the consistent global trajectories.

In this thesis, we first propose a decentralized relative state estimation for aerial swarms based on visual-inertial-UWB fusion to address the relative localization problem of the aerial swarm. This algorithm is improved in subsequent studies, including improving the initialization problem, introducing omnidirectional cameras to solve the observability problem, and introducing map-based localization to feature global consistency.

We also investigate distributed SLAM for aerial swarms to solve the scaling issue. BDPGO, a distribute...[ Read more ]

In recent years, aerial swarm technology has developed rapidly. In order to accomplish a fully autonomous aerial swarm, a key technology is collaborative SLAM (CSLAM) for aerial swarms, which estimates the relative pose and the consistent global trajectories.

In this thesis, we first propose a decentralized relative state estimation for aerial swarms based on visual-inertial-UWB fusion to address the relative localization problem of the aerial swarm. This algorithm is improved in subsequent studies, including improving the initialization problem, introducing omnidirectional cameras to solve the observability problem, and introducing map-based localization to feature global consistency.

We also investigate distributed SLAM for aerial swarms to solve the scaling issue. BDPGO, a distributed pose graph optimization (DPGO) algorithm that can schedule computational resources on different unmanned aerial vehicles (UAV) to solve distributed optimization in a balanced way, is proposed in this thesis. Our evaluation shows that BDPGO improves the solving speed and reduces the communication overhead compared to the previous DPGO methods.

Finally, based on the experience accumulated from our previous research, we propose D²SLAM: a decentralized and distributed (D²) collaborative SLAM algorithm. This algorithm has high local accuracy and global consistency, and the distributed architecture allows it to scale up. D²SLAM covers swarm state estimation in two scenarios: near-field state estimation for high real-time accuracy at close range and far-field state estimation for globally consistent trajectories estimation at the long-range between UAVs.

Our proposed methods successfully demonstrate centimeter-level accuracy on datasets and extensive experiments. With the proposed methods, we have successfully conducted extensive realistic experiments, including inter-UAV collision avoidance and multi-UAV unknown environment exploration.

The techniques proposed in this thesis will contribute to the aerial swarm toward full autonomy and efficiency.