A3Framework: A Cloud Framework with Autonomous Driving Path Planning

This paper proposes an autonomous navigation cloud framework, the A3Framework, which enhances real-time performance through an Edge-Cloud system. To adapt to the cloud platform's enhanced real-time performance, global planning uses highly efficient A * and ACO algorithms in terms of runtime efficiency and memory usage. To improve path quality and reduce communication time costs with the vehicle, an improved APF algorithm with methods such as step caching is used in local planning. Multiple sensors are used for dynamic mapping. Experimental results and theoretical analysis show that this framework significantly improves path performance metrics compared to traditional algorithms, such as travel time. Moreover, its real-time performance is greatly enhanced when deployed to a cloud platform.