Socializing A* Algorithm for Crowd- and Socially Aware Navigation
Abstract
Today, an undeniable interest is given to the development of socially intelligent
robotic systems and efficient crowd- and socially-aware navigation strategies. In
this paper, we introduce a novel crowd-aware navigation algorithm that combines
the A* path planner with the Double Deep Q-Network (DDQN) method.
The algorithm, named Social A*, aims to allow safe navigation for mobile robots
in social dynamics and crowded environments. In order to facilitate future realworld
implementation, a new learning environment compatible with the Robot
Operating System (ROS) is developed. This allows expert teleoperation to help
train the DDQN agent and refine the reward function. We conducted extensive
simulations to compare the performance of Social A* with Socially-attentive
Reinforcement Learning (SARL*) and Intention Aware Robot Crowd Navigation
with Attention-Based Interaction Graph (IARL). The obtained simulation results
demonstrate that Social A* not only surpasses SARL* and IARL but also shows
enhanced performance in handling static obstacles. These results showcase the
excellent crowd-aware navigation performance, the efficiency, and the significant
potential of the algorithm.
Origin | Files produced by the author(s) |
---|