Design And Implementation Of Service Robot Remote Procedure Calling Framework

In recent years,with the development of robot technology,service robots have gradually entered people's daily lives.People have put forward higher requirements for the service robots to be intelligent,hoping that service robots can provide more targeted services based on human expressions.However,the mainstream expression detection algorithms are based on deep learning,which requires a large amount of calculation.The computing power of the robot side is limited.If expression detection is performed on the robot side,it is difficult to ensure the real-time detection,which will also affect the robot's processing of other tasks and the robot's battery life.At present,communication technology is developing rapidly,so we consider putting complex and time-consuming expression detection algorithms on the server side,and the robot side calls the server-side algorithm through the network to save computing resources on the robot side.We design and implements a remote procedure call framework that allows the robot to call server-side algorithms like local functions.In order to ensure the high availability and scalability of the system,the framework can increase or decrease the computing power of the server side without the perception of the robot side.If several hosts on the server side fail,it will not affect the overall operation of the system.We use Protocol Buffers as the intermediate description language of the remote procedure call framework,uses HTTP / 2 based g RPC that supports two-way streaming data transmission to implement remote procedure calls,and uses Rabbit MQ message queues to distribute the requests,ensuring the system keeping available when the number of requests is large and uses Zookeeper for distributed coordination to achieve the sharing of configuration information among multiple hosts and the management of multiple instances of g RPC.Finally,we use Kubernetes for cluster management to achieve resource scheduling and dynamic scaling of services.Experiments show that the our framework can save 60% of the traffic compared with the commonly used RESTful framework,and it has obvious advantages when the network bandwidth is low,and it is more robust when the network has delays and packet loss.It is more suitable for occasions with high requirements on real-time and reliability.