Abstract

This work presents advancements achieved in the field of autonomous service robotics in open accessible spaces, with preliminary project results. A novel navigation algorithm is presented. It integrates (i) a new trajectory planner which can be proven to be safe with respect to expected moving objects by making use of emergency trajectories, (ii) person motion models trained from historical data and integrated into the trajectory planner to mimic human-like behaviors, and (iii) a global planner which is able to coordinate multiple vehicles in tight spaces without handcrafted tracks to prevent deadlocks. The results presented are shown on real and simulated hardware which was developed and used in the projects TransitBuddy, TransportBuddy and Autonomous Fleet.