A1 SLAM: Quadruped SLAM using the A1's Onboard Sensors

Quadrupeds are highly versatile robots that can traverse over difficult terrain that wheeled mobile robots are unable to. This flexibility makes quadrupeds appealing for various applications, such as inspection, surveying construction sites, and search-and-rescue. However, to effectively perform these tasks autonomously, quadrupeds, as with other mobile robots, require a form of perception that will enable them to localize when placed in an environment without a priori knowledge. For robots to know its location in the environment, it must localize against a predefined map, but a robot can only create a map based on its known location. To solve this chicken-and-egg problem, simultaneous localization and mapping, or SLAM, is the standard approach used for mobile robots by optimizing for the robot's location and map simultaneously. The estimated poses and map from SLAM algorithms can then be used for downstream tasks such as facilitating controllers depending on the terrain or planning in navigation. Despite the recent developments in both quadruped robotics and in SLAM research, there has yet to be an open-source package that is specifically designed for high performing SLAM on quadrupeds.