A prediction based algorithm for dynamic load balancing is proposed, which is capable of predicting the time for a node to reach the idle state based on the local loading information and issuing a task request before this so as to ensure that all nodes in the system will remain busy to enhance the utilization rate of the system resources and the overall system performance.

This design is made for an embedded system and has been successfully implemented In the scheme, the interface functions first accept the request task, then the execution level function which runs as a thread sends out request message and receives and parses response message.

In this model,service requests with dynamic priorities are sorted through three-class multi-heap multithreads,consequently the model conducted the orderly service functions according to the priority of request.

Firstly, according to the number of data items of requests and the process of repeated access for data items, some multi-item requests are found out.

On the condition of meeting the grasping stability and the task requirement,the optimal grasping configuration is determined with the goal of minimizing of the maximal finger contact forces during the entire manipulating process.