Qi Zhang, Weidong Tang, and Meiling Liu


Robotic technology, performance optimisation, probabilistic model checking, behaviour-centric algorithm, Markov decision processes


With the continuous development of robotics, more and more intelligent robot products are used in different fields. How to optimise the overall performance of the robot so that the system can complete its tasks more safely, reliably, and efficiently has become the key to developing robotics. To address the problem of overall robot performance optimisation, this paper proposes a probabilistic model checking method for robot performance optimisation. The approach begins by abstracting the general behaviour and system components of the robot when interacting with the environment. Then construct the algorithm for the interactive run of each component of the system centered on robot behaviour, and using behavioural algorithms in conjunction with Markov decision processes (MDP) to construct probabilistic formal models through the PRISM language. Finally, the results are analysed and validated using probabilistic model checking. This method is applicable to all robotic systems that can abstract the behaviour and system components. The feasibility of the method is verified through two scenarios given in the paper, and the overall performance of the robot is analysed and optimised in terms of task completion time, optimal path, power consumption of the robot, failures, etc., while safeguarding system safety and reliability, and the correctness and effectiveness of the method for robot performance optimisation are demonstrated through experiments.

Important Links:

Go Back